# Bearly Documentation — Full

> Complete consumer documentation for Bearly in one Markdown document.

Index: [https://bearly.ai/docs.md](https://bearly.ai/docs.md)

## Documentation directory

### Getting Started

- [Get started with Bearly](https://bearly.ai/docs.md): Sign in, start a conversation, and learn the essential controls for working with Bearly.
- [Have your first conversation](https://bearly.ai/docs/chats/first-chat.md): Start a Bearly chat, guide the answer, revise your request, and use the controls on each message.
- [Work with files in a chat](https://bearly.ai/docs/chats/files.md): Attach files, follow upload progress, preview results, and download files from a Bearly conversation.
- [Share a conversation or file as a Page](https://bearly.ai/docs/chats/share.md): Create a standalone Page from your work, review it, choose who can open it, and share a link that others can continue from in Bearly.
- [Organize your chats](https://bearly.ai/docs/chats/organize.md): Search, rename, pin, archive, move, select, and delete Bearly conversations.
- [Ask Bearly to manage your chats](https://bearly.ai/docs/chats/manage-with-bearly.md): Find, read, create, rename, move, and delete conversations by asking Bearly in chat.

### Research & Context

- [Search the web, find images, and run Deep Research](https://bearly.ai/docs/features/web-search-deep-research.md): Find current information and visual references in chat, or plan a longer investigation that produces a sourced report.
- [Search academic papers with Scholar](https://bearly.ai/docs/features/scholar.md): Ask a research question and get an answer built on papers, with each claim cited to a passage you can open and check.
- [Watch and check YouTube videos](https://bearly.ai/docs/features/youtube.md): Paste a YouTube link into a chat to watch the video with a verdict, key moments, shorter cuts, and a summary, look up what it mentions, and check the figures it states.
- [Look up a person with People Search](https://bearly.ai/docs/features/people-search.md): Confirm which person you mean, then get a cited brief of their public work with a source and date on every claim.
- [Research your documents](https://bearly.ai/docs/features/document-research.md): Ask focused questions across attached or project files, compare sources, and open cited passages for verification.
- [Summarize a document](https://bearly.ai/docs/features/deep-summary.md): Create a structured Deep Summary of a long document, then continue working with it in the same chat.
- [Find conversations and control memories](https://bearly.ai/docs/features/memory-search.md): Search your Bearly chat history and choose which earlier conversations a chat can use for context.
- [Work with local files](https://bearly.ai/docs/features/work-locally.md): Let Bearly read, create, edit, and run files on your computer from a folder you choose.
- [Use your computer with Bearly](https://bearly.ai/docs/features/computer-use.md): Let Bearly complete tasks in visible desktop apps while you control what it can see, do, and how long access lasts.
- [Browse beside a Bearly chat](https://bearly.ai/docs/features/browser.md): Open sites in Bearly Browser, annotate a page, or approve limited site access for a chat task.
- [Use Bearly on a web page](https://bearly.ai/docs/features/browser-extension.md): Summarize, question, or explain the page you are reading with the Bearly extension for Chrome, without leaving the tab.
- [Control Chrome from Bearly](https://bearly.ai/docs/features/chrome-control.md): Let a Bearly desktop chat read, navigate, and inspect your regular Chrome tabs, including sites you are already signed in to.
- [Find and check domain names](https://bearly.ai/docs/features/domain-research.md): Ask Bearly for domain ideas around a product, company, or theme, then check whether a specific domain is still available to register.
- [Research keywords, rankings, and competitors with SEO Data](https://bearly.ai/docs/features/seo-data.md): Pull keyword demand, live Google results, a site's search visibility, competitor gaps, backlinks, or a page audit into a chat, then keep working with the evidence.
- [Find places and plan routes with Maps](https://bearly.ai/docs/features/maps.md): Ask Bearly about places, reviews, directions, and travel times, and get cards with ratings, hours, routes, and stops along the way.
- [Check weather, alerts, air quality, and pollen](https://bearly.ai/docs/features/weather.md): Ask Bearly for current conditions, forecasts, public weather alerts, air quality, pollen, or past weather for a place, with results shown as cards in the chat.

### Create & Transform

- [Create interactive AI artifacts with Pages](https://bearly.ai/docs/features/artifacts.md): Build, refine, reopen, and download interactive AI artifacts such as calculators, dashboards, planners, and small tools in Bearly.
- [Edit a long draft in Canvas](https://bearly.ai/docs/features/canvas.md): Turn a Bearly response into a Markdown draft, edit it directly, preview the formatting, and ask AI for focused revisions.
- [Analyze data with Code Interpreter](https://bearly.ai/docs/features/code-interpreter.md): Let Bearly run Python for calculations, data analysis, charts, file conversion, and generated files.
- [Create charts from spreadsheets](https://bearly.ai/docs/features/charts.md): Turn a spreadsheet or CSV into a KPI card, bar chart, or line chart with the source data ready to inspect and download.
- [Create and edit documents](https://bearly.ai/docs/features/document-editor.md): Use Document Editor to create, revise, and convert PDF, Word, PowerPoint, and Excel files in a Bearly chat.
- [Check grammar and clarity](https://bearly.ai/docs/features/grammar-correction.md): Review grammar, spelling, punctuation, style, vocabulary, and clarity suggestions without losing your original meaning.
- [Generate images](https://bearly.ai/docs/features/image-generation.md): Create images in chat, try models in parallel, and keep editing from any result.
- [Edit an image](https://bearly.ai/docs/features/image-editing.md): Change an uploaded or generated image, select an area to edit, or extend its edges in the conversation.
- [Combine images](https://bearly.ai/docs/features/image-combining.md): Select multiple source images, describe how their elements should relate, and create a new composition in chat.
- [Create and edit videos](https://bearly.ai/docs/features/video-generation.md): Generate a short video from text or an image, request sound and format choices, and transform an existing clip.
- [Download a video](https://bearly.ai/docs/features/download-video.md): Save videos from supported public sites to a Bearly chat, follow download progress, and keep the resulting files.
- [Turn text into speech](https://bearly.ai/docs/features/text-to-speech.md): Listen to a Bearly answer or create downloadable spoken audio with the voice and delivery you want.
- [Transcribe audio and video](https://bearly.ai/docs/features/transcripts.md): Turn recordings into searchable text, follow along with playback, label speakers, and download subtitles.

### Workflows

- [Run prompts on a schedule with Routines](https://bearly.ai/docs/features/routines.md): Create a routine that runs a saved prompt on a schedule, keeps every run in its own chat, and can email or text you the result.
- [Record and transcribe meetings](https://bearly.ai/docs/features/meetings.md): Record online meetings, in-person conversations, and voice notes in Bearly, with a live transcript, speaker names, summaries, and a chat that starts from the transcript.
- [Dictate instead of typing](https://bearly.ai/docs/features/dictation.md): Turn on dictation, speak into the message box with a click or a held key, and choose what happens to the finished transcript.
- [Talk with Bearly by voice](https://bearly.ai/docs/features/voice-conversations.md): Start a live voice conversation, manage the microphone and call, and continue from the saved transcript.
- [Add workflows with Skills](https://bearly.ai/docs/features/skills.md): Install, manage, and use reusable Bearly workflows from the Skill Store or a trusted source.
- [Create reusable Prompt Templates](https://bearly.ai/docs/features/prompt-templates.md): Turn a recurring instruction into a reusable form that you can fill out and submit in any Bearly chat.
- [Compare answers with Council of Experts](https://bearly.ai/docs/features/council-of-experts.md): Ask several AI models the same question, inspect their individual answers, and use Bearly's synthesis.
- [Send email or text messages from a chat](https://bearly.ai/docs/features/send-email.md): Send formatted email or a concise text message from a Bearly chat, then confirm the result and recover from failures.
- [Connect services and MCP servers](https://bearly.ai/docs/features/connectors.md): Add a supported service or your own MCP server, then choose which chats and Routines can use its tools.

### Projects

- [Organize chats with projects](https://bearly.ai/docs/projects/overview.md): Keep related chats, files, instructions, and settings together in a Bearly project.
- [Add files to a project](https://bearly.ai/docs/projects/documents.md): Upload, search, inspect, and manage files that Bearly can use across a project's chats.
- [Set project-wide chat defaults](https://bearly.ai/docs/projects/settings.md): Choose the model, language, style, instructions, and document behavior for chats in a project.
- [Share a project with your team](https://bearly.ai/docs/projects/collaboration.md): Give a team or individual teammates Reader, Writer, or Owner access to a Bearly project.
- [Protect a project with a password](https://bearly.ai/docs/projects/encryption.md): Add, use, change, or remove password protection for a Bearly project.

### Organizations

- [Create and manage an organization](https://bearly.ai/docs/teams/overview.md): Create a Bearly organization, review its account details, rename it, leave it, or delete it when eligible.
- [Set an organization access banner](https://bearly.ai/docs/teams/access-banner.md): Require organization members to accept a custom message before using Bearly.
- [Manage organization members](https://bearly.ai/docs/teams/members.md): Add people to a Bearly organization, import a roster from CSV, change member access, and remove members.
- [Manage groups and group admins](https://bearly.ai/docs/teams/groups.md): Organize departments within a Bearly organization, delegate group management, set policies, and review shared projects and usage.
- [Connect Microsoft Entra to an organization](https://bearly.ai/docs/teams/microsoft-entra.md): Let a managed Bearly organization sign in with Microsoft work or school accounts, and provision members and groups from Microsoft Entra.
- [Manage organization billing and seats](https://bearly.ai/docs/teams/billing.md): Start an organization billing account, buy and assign seats, change seat plans, and resolve payment issues.
- [Set organization policies](https://bearly.ai/docs/teams/policies.md): Set shared chat defaults, model access, feature permissions, and spending limits for members of a managed organization.
- [Review organization usage](https://bearly.ai/docs/teams/usage.md): Compare a managed organization's AI spend and calls across members, models, and recent months.
- [Distribute skills from a shared folder](https://bearly.ai/docs/teams/skill-stores.md): Register a managed organization's shared skill library and assign individual skills through organization policies.
- [Review organization activity](https://bearly.ai/docs/teams/activity.md): See who changed your organization's settings, policies, and membership, with readable before-and-after details.

### Settings & Account

- [Change your chat settings](https://bearly.ai/docs/settings/chat-settings.md): Choose how Bearly responds in one conversation, set defaults for new chats, and control the personal context shared with AI models.
- [Choose how Bearly uses external tools](https://bearly.ai/docs/settings/agent-permissions.md): Choose Standard or Full access for chats and routines, review saved access, and manage organization eligibility.
- [Manage billing and review usage](https://bearly.ai/docs/account/billing-usage.md): Change your Bearly plan, review the current credit period, add a credit boost, and see monthly usage by model.
- [Cancel your Bearly subscription](https://bearly.ai/docs/account/cancel-subscription.md): Stop renewal for a Bearly subscription purchased on the web or through Apple's App Store, and confirm the cancellation date.
- [Delete your Bearly account](https://bearly.ai/docs/account/delete-account.md): Permanently delete your Bearly account, private work, credentials, and local Bearly data after reviewing subscriptions and retained billing records.

### Privacy & Security

- [Privacy and security in Bearly](https://bearly.ai/docs/privacy-security.md): Understand how private AI works in Bearly, what a model receives, and which privacy, memory, encryption, and access controls you can use.
- [Use zero-data-retention models](https://bearly.ai/docs/zero-data-retention.md): See which models run with zero data retention, restrict Bearly to those models, and understand what the setting covers.
- [Turn on unitemized billing](https://bearly.ai/docs/unitemized-billing.md): Turn on unitemized billing so your usage history keeps one rounded total per day instead of a line for every request.

### Changelog

- [Changelog](https://bearly.ai/docs/changelog.md): Every Bearly release since June 2024, newest first, with what changed in each one.

### Legal

- [Privacy Policy](https://bearly.ai/privacy-policy.md): Bearly AI Privacy Policy
- [Terms of Service](https://bearly.ai/terms-of-service.md): Bearly AI Terms of Service

---

---
title: "Get started with Bearly"
description: "Sign in, start a conversation, and learn the essential controls for working with Bearly."
canonical: "https://bearly.ai/docs/"
markdown: "https://bearly.ai/docs.md"
---

# Get started with Bearly

HTML: [https://bearly.ai/docs/](https://bearly.ai/docs/)

> Sign in, start a conversation, and learn the essential controls for working with Bearly.

Bearly gives you one place to ask questions, work with files, research ideas, and return to earlier conversations.

## Choose where to work

- **Web:** open [Bearly in your browser](https://bearly.ai/dashboard/focus). The web app works without an installation.
- **Desktop:** visit [bearly.ai](https://bearly.ai) and choose the download for Mac, Windows, or Linux. If you are already signed in on the web, open **App Settings** → **Get Desktop App**.
- **iPhone or iPad:** install [Bearly from the App Store](https://apps.apple.com/us/app/bearly-ai/id6445975220). Other mobile devices can use the web app.

Bearly uses the same account across its apps. A chat protected to one device must still be opened on that device, and features that work with local files or other apps may require Bearly for desktop.

## Sign in



1. Open Bearly and select **Continue with Google** or **Continue with Microsoft** when it is offered, or enter your address under **Email** and select **Continue**. Microsoft sign-in works for organizations that have [connected Microsoft Entra to their Bearly team](https://bearly.ai/docs/teams/microsoft-entra.md).
2. If you continue by email, open the message from Bearly. Paste the three-word code—or type one word into each field—under **Check your email**.
3. Select **Sign in**.



If Bearly says the codewords are invalid, select **Back**, confirm the email address, and request a new code. Some managed accounts require a work email.

## Start your first chat



1. Select **New Chat**.
2. Write what you want in the message box. A useful first request includes the outcome, any context Bearly should know, and the form you want back.

   For example: `Turn these meeting notes into a concise decision memo with owners and open questions.`

3. To include a file, select the **+** button and choose **Attach files**. Wait for the file to finish uploading or analyzing.
4. Select **Send message**, or press <kbd>Enter</kbd> on a computer.



Bearly begins answering in the same chat. You can keep asking follow-up questions without repeating the whole request. Select **Stop** if you need to end an answer while it is being written.

> **Tip — Your work is organized as you go**
> Bearly gives an active conversation a title and places it in the sidebar. You can rename, pin, archive, move, or delete it later.


## Where to go next

- [Have your first conversation](https://bearly.ai/docs/chats/first-chat.md) for editing, rerunning, copying, and deleting messages.
- [Work with files in a chat](https://bearly.ai/docs/chats/files.md) for previews, downloads, and the file browser.
- [Organize your chats](https://bearly.ai/docs/chats/organize.md) for search, projects, archives, and bulk actions.
- [Create a project](https://bearly.ai/docs/projects/overview.md) when several chats and documents belong to the same body of work.

If Bearly cannot prepare a new chat, it keeps your draft in the composer and usually retries automatically. Use **Try again** if the notice remains after your connection returns. For other problems, contact [Bearly Support](mailto:support@bearly.ai).

---

---
title: "Have your first conversation"
description: "Start a Bearly chat, guide the answer, revise your request, and use the controls on each message."
canonical: "https://bearly.ai/docs/chats/first-chat/"
markdown: "https://bearly.ai/docs/chats/first-chat.md"
---

# Have your first conversation

HTML: [https://bearly.ai/docs/chats/first-chat/](https://bearly.ai/docs/chats/first-chat/)

> Start a Bearly chat, guide the answer, revise your request, and use the controls on each message.

A Bearly chat is a continuing workspace: begin with a concrete request, inspect the answer, and refine it with follow-up messages.

## Send a request



1. Select **New Chat** in the sidebar.
2. If the work belongs to a project, select the project name in the context strip before sending the first message. A chat cannot be reassigned from this control after it begins, but you can [move it later](https://bearly.ai/docs/chats/organize.md#move-chats-into-projects).
3. Write your request in the composer. Include the result you want, the relevant context, and any constraints that would change the answer.

   For example: `Compare these two proposals in a table, then recommend one. Favor lower operational risk over short-term savings.`

4. Select **Send message**, or press <kbd>Enter</kbd> on a computer.



Bearly shows your request followed by its answer. Continue in the same composer to ask for a revision, challenge an assumption, or add information.

## Stop or redirect an answer

While Bearly is working, **Send message** becomes **Stop**. Select it to end the current answer. Anything already shown remains in the conversation.

If you type a new message while Bearly is still working, the control changes to **Interrupt**. Select it to stop the current work and send the new direction.

Stopping does not undo actions that already finished. If a tool asks for approval, read the request before deciding whether to allow it.

## Revise an earlier request

On a computer, hover over your message to reveal its controls. On a phone or tablet, the controls remain visible beneath it.



1. Select **Edit Query** under the request you want to change.
2. Edit the text or attachments in the composer. The context strip reads **Editing message**.
3. Select **Send message** to run the revised request, or select **Cancel** to leave the original unchanged.



The revision becomes another version of that point in the conversation. Use **Previous version of query** and **Next version of query** to move between versions.

## Rerun a request

When model controls are available, use **Rerun with same model** beneath a request when you want a fresh answer without changing the wording. To try another available model, open the model name beside **Rerun** and choose one from the menu.

Available models can depend on your plan and, for managed accounts, your team's policy. A model that is unavailable may prompt you to change plans or choose another option.

## Copy or remove a message

- Select **Copy** beneath either side of an exchange to copy that request or answer with its formatting.
- Select **Query info** to see when the exchange started and completed, along with its duration.
- Select **Delete Query**, then confirm **Delete Message**, to remove the request and its answer from the chat.

> **Danger — Deleted messages cannot be restored**
> Deleting a query removes that exchange immediately. To remove the entire conversation, use the chat menu in the sidebar instead.


## If an answer fails

- If a message says **Unable to load message**, select **Get help** to report the rendering problem.
- If Bearly stops because of a temporary connection or service problem, rerun the request after the connection recovers.
- If a new chat cannot be prepared, keep the page open. Bearly preserves the draft, retries temporary failures, and shows **Try again** when a manual retry is useful.

To add material to the conversation, see [Work with files in a chat](https://bearly.ai/docs/chats/files.md). To find or tidy earlier work, see [Organize your chats](https://bearly.ai/docs/chats/organize.md).

---

---
title: "Work with files in a chat"
description: "Attach files, follow upload progress, preview results, and download files from a Bearly conversation."
canonical: "https://bearly.ai/docs/chats/files/"
markdown: "https://bearly.ai/docs/chats/files.md"
---

# Work with files in a chat

HTML: [https://bearly.ai/docs/chats/files/](https://bearly.ai/docs/chats/files/)

> Attach files, follow upload progress, preview results, and download files from a Bearly conversation.

Attach source material to a request, or open and download files that Bearly creates while you work.

## Attach a file

Each attachment must be 300 MB or smaller. Whether Bearly can read and preview a file depends on its format; common documents, text and data files, images, audio, and video are supported.



1. Open a chat or select **New Chat**.
2. Select the **+** button in the composer, then choose **Attach files**.
3. Choose one or more files. You can also drag files onto the chat. On a phone or tablet, **Take photo** appears in the **+** menu when the camera is available.
4. Wait while the attachment shows **Uploading** and, when needed, **Analyzing**. The file is ready when those indicators disappear.
5. Add a request that says what Bearly should do with the material, then select **Send message**.



For example: `Read the attached report. Give me the three decisions it supports, the evidence for each, and any unanswered questions.`

To remove an attachment before sending, select **Remove file** on its preview. If it shows **An error occurred**, remove it and attach it again after checking your connection and the 300 MB limit.

> **Note — Files belong to a particular context**
> A file attached in the composer is an **Uploaded** file in that chat. A document added to a project is a **Project** file and is available to chats in that project. A file Bearly creates is a **Generated** file. These labels help you tell the sources apart; they are not interchangeable copies.


## Preview a file

Select a file in the conversation to open its preview. Bearly chooses a view suited to the format:

- Images, PDFs, audio, video, Markdown, text, CSV, JSON, and many code files can be viewed directly.
- Word, PowerPoint, Excel, and similar documents can open as a PDF preview when conversion is available.
- An unfamiliar file type opens a simple file card instead. You can still open it externally or download the original.

The preview opens beside the chat on larger screens and takes over the workspace on smaller ones. Select **Close** to return to the conversation. On iPhone or iPad, PDFs open outside the in-app preview.

## Find every file in the current chat

When the current chat has files, its header shows the **Browse files** button with a file count.

1. Select **Browse files**.
2. Choose **All**, **Project**, **Uploaded**, or **Generated** to narrow the list. **Project** appears only when the chat belongs to a project.
3. Select a filename to open it, or select **Download** on its row to save it directly.

An empty filter shows **No files in this category**. If the **Browse files** button is absent, the chat does not yet have any available files.

## Download or copy a result

Open a file and use the controls in its header:

- **Download** saves the file in its original format.
- When an Office document has a PDF preview, the download menu offers **PDF** and **Original**.
- **Copy** is available for text, code, data, and images that Bearly can copy directly.
- **Open externally** hands the file to the appropriate browser or app.

If a preview cannot load, download the original and open it in its usual application. A failed conversion does not replace or alter the original file.

For long-lived reference material shared by several conversations, [add documents to a project](https://bearly.ai/docs/projects/documents.md) instead of attaching the same file repeatedly.

To send a file to someone outside the conversation, [share it as a Page](https://bearly.ai/docs/chats/share.md) with a download link and your chosen audience.

---

---
title: "Share a conversation or file as a Page"
description: "Create a standalone Page from your work, review it, choose who can open it, and share a link that others can continue from in Bearly."
canonical: "https://bearly.ai/docs/chats/share/"
markdown: "https://bearly.ai/docs/chats/share.md"
---

# Share a conversation or file as a Page

HTML: [https://bearly.ai/docs/chats/share/](https://bearly.ai/docs/chats/share/)

> Create a standalone Page from your work, review it, choose who can open it, and share a link that others can continue from in Bearly.

Sharing turns the substance of a conversation into a standalone Page with artwork and a link preview. Your full conversation is not shared. You review the Page before publishing it.

You can also share a file from the chat. Preparing a share uses text and image generation, which counts toward your Bearly usage. Sharing must be allowed by your organization, and the conversation must allow changes.

## Prepare a conversation to share



1. **Open the conversation and select Share**

   Use the **Share** icon in the chat header, beside **Browse files** when the chat has files. Select **Prepare share**.

   You can also ask Bearly directly: `Prepare a shareable Page of the final proposal in this chat.` Name the result you want when the conversation contains several drafts or topics.

2. **Watch the Page take shape**

   The **Preparing share** screen shows the title first, then builds the artwork and Page together. It shows separate progress for the Page and its artwork.

   If the first image model cannot make the artwork, Bearly tries another. If both are unavailable, it uses a simple illustration so you can still review and publish the Page. Check the link preview before publishing.

   Select **Stop building** to cancel preparation. Closing the screen with **Back to chat** lets preparation continue; select **Share** again to return to it. Artwork is saved only when Bearly has a complete Page to prepare. Any files already saved remain in the chat.

3. **Review the finished Page**

   The Page opens with its **Ready to share** controls. Check the Page itself and the title, description, and artwork in the link preview. Expand **Context for a new conversation** to review the brief that a reader can bring into a new chat.

   If the content needs a change, ask Bearly to revise it and prepare the updated version before publishing. For a story, the Page should contain the story itself. For a proposal, it should contain the proposal and the context its reader needs.



Preparation does not publish anything. The Page is ready to send only after you complete the publishing step below.

## Publish and send the link



1. **Choose Audience**

   - **Anyone with the link** lets people open the Page without signing in. Anyone they forward the link to can also open it.
   - **People and teams** requires readers to sign in with an account that has access. After publishing, use **Manage viewers** to add the people or teams who should be able to open it.

   Your organization's policy determines which choices appear.

2. **Select Publish**

   When publication succeeds, the controls show the Page's link. For **People and teams**, open **Manage viewers**, search for a teammate by name or email, and select **Add**, or add one of the listed teams.

3. **Copy or share**

   Select **Copy link**. The button briefly shows **Copied**. **Share…** opens your device's share sheet when it is available.



Public links include a title, description, and artwork for apps that display link previews. Restricted links keep those details behind sign-in, so they may show a generic preview in another app.

## Share a file

Open the file from the conversation or **Browse files**, then select its **Share file** icon. Select **Prepare share**, review the Page and the included filename, and publish it using the same audience controls.

An existing Page keeps its content. When sharing another kind of file, Bearly creates a Page with a download of the original. Images, PDFs, audio, and video can also appear in the Page when the reader's browser supports viewing or playing them. If the preview is unavailable, use the download link.

Each share can include one original file up to 300 MB. A Page you prepare for sharing must be under 9 MB. Files included in a share use the Page's audience.

## Continue from a shared Page

On a published Page, select **Continue conversation in Bearly** and sign in if prompted. Bearly opens a new chat with the Page's link and a short brief attached. Add your question, then send the message when you are ready.

This starts your own conversation. It does not copy the author's full thread or import the original files as editable attachments. Download a shared file and [attach it to your chat](https://bearly.ai/docs/chats/files.md) if you want Bearly to work on that file.

## Update or stop sharing

Reopen the shared Page in the original chat and select **Publish** to manage it.

- **Publish update** replaces the published version at the same link. Changes made in the chat do not update the public Page until you publish them.
- **Audience** changes who can open the link. For restricted links, use **Manage viewers** → **Remove** to remove a person or team.
- **Unpublish** stops access through the shared link. A cached public Page may remain visible briefly, and unpublishing cannot recall copies someone has already downloaded or saved elsewhere.

## If sharing does not finish

If preparation stops or fails, the draft screen keeps the content received so far and says what failed. Select **Prepare again** when you want to start another attempt; generation uses your Bearly usage again. If the Page could not be read, check that the source file is still available. If saving failed, check your connection before trying again.

If Page details fail to load, select **Retry**. If publishing fails, keep the saved Page and retry from **Publish**. If unpublishing shows **Retry cleanup**, select it to finish cleanup.

If a reader sees **You need access**, check the Page's audience and viewers, and ask them to sign in with the account you added. If sharing controls are unavailable, check whether the chat is read-only or ask your organization admin about its [sharing policy](https://bearly.ai/docs/teams/policies.md).

---

---
title: "Organize your chats"
description: "Search, rename, pin, archive, move, select, and delete Bearly conversations."
canonical: "https://bearly.ai/docs/chats/organize/"
markdown: "https://bearly.ai/docs/chats/organize.md"
---

# Organize your chats

HTML: [https://bearly.ai/docs/chats/organize/](https://bearly.ai/docs/chats/organize/)

> Search, rename, pin, archive, move, select, and delete Bearly conversations.

Use the sidebar to keep active work close, move related conversations into projects, and clear away chats you no longer need.

## Find an earlier conversation

Select the **Search conversations** magnifying glass in the sidebar header. On a computer, you can also press <kbd>Command</kbd>+<kbd>F</kbd> on macOS or <kbd>Ctrl</kbd>+<kbd>F</kbd> on Windows and Linux.

Enter a title, phrase, or topic under **Search conversations...**. Results show the matching conversation, a short excerpt, and the number of matches. Select a result to open that chat.

Search covers conversation titles and contents. If nothing matches, Bearly shows **No results found**; try a shorter or more distinctive phrase.

## Rename or pin a chat

Open the chat's three-dot menu in the sidebar:

- Select **Rename**, enter the new title, then press <kbd>Enter</kbd> or click elsewhere to save it.
- Select **Pin** to place the conversation under **Pinned**. Open the same menu and select **Unpin** when it no longer needs special prominence.
- Select **Mark unread** to give the conversation an unread marker. Opening it marks it read, or you can choose **Mark read** from the menu.

You can also rename the open conversation by selecting its title in the chat header.

## Move chats into projects

**Move to project...** appears after you have at least one project.



1. Open a chat's three-dot menu and select **Move to project...**.
2. Search for a project if needed, then select its name. A project already containing every selected chat is marked **Current**.
3. To move a chat back to the main list, open the menu again, choose **Move to project...**, then select **Remove from project**.



On a computer, you can also drag a chat onto a project in the sidebar. The project controls which project documents and settings are available to that conversation.

## Archive finished work

Open a chat's three-dot menu and select **Archive**. The chat leaves the main list but is not deleted.

Select **Archives** near the sidebar footer to see everything you have archived in one list. Archived chats and projects are grouped by when you archived them, newest first, and each one shows its archive date. A chat that belongs to a project also shows the project's name.

Once the list holds eight or more items, a **Filter archives** box appears above it. Enter part of a chat title or project name to narrow the list.

Open an archived chat's menu and select **Unarchive** to return it to the active list. To leave the list without changing anything, select the **Back to all chats** arrow or select **Archives** again. When you unarchive the last item, the list closes on its own.

Archived chats cannot be deleted. Unarchive one before choosing **Delete**.

## Act on several chats



1. Open any chat's three-dot menu and select **Select**.
2. Use the checkboxes to choose more conversations. Choose **Select all** only when you want every chat in the current main list or active filtered view.
3. Select **Move**, **Archive**, or **Delete**. Select **Cancel** to leave selection mode without changing anything.



In the **Archives** list, **Unarchive** takes the place of **Archive** and restores every selected chat at once. **Delete** does not remove archived chats there; Bearly asks you to unarchive them first.

## Delete a chat

Open the chat's three-dot menu, select **Delete**, review the confirmation, then select **Delete** again.

> **Danger — Deleting a chat is permanent**
> The conversation, its attached files, and its chat memory are deleted and cannot be restored. Archive the chat if you merely want it out of the active list.


Deleting one message is different from deleting the chat. Use **Delete Query** beneath a message when you want to remove only that exchange.

For a durable workspace around several related conversations and files, see [Create and use projects](https://bearly.ai/docs/projects/overview.md).

---

---
title: "Ask Bearly to manage your chats"
description: "Find, read, create, rename, move, and delete conversations by asking Bearly in chat."
canonical: "https://bearly.ai/docs/chats/manage-with-bearly/"
markdown: "https://bearly.ai/docs/chats/manage-with-bearly.md"
---

# Ask Bearly to manage your chats

HTML: [https://bearly.ai/docs/chats/manage-with-bearly/](https://bearly.ai/docs/chats/manage-with-bearly/)

> Find, read, create, rename, move, and delete conversations by asking Bearly in chat.

From any Bearly chat, you can ask Bearly to find and organize other conversations without leaving the chat you are using.

## Find and read an earlier chat

Describe the work you remember and what you need from it:

```text copy
Find the conversation where we agreed on the onboarding milestones and summarize the final plan.
```

Bearly searches the conversation memory available to the current chat, identifies likely matches, and reads the relevant conversation before answering. If a conversation is large, Bearly may add its complete contents to the current chat as a Markdown file and read the file in parts.

Topic search follows the current chat's **Memories** selection. In a project chat, **Project Only** searches that project's conversations and **All Chats** searches more broadly. If Bearly cannot find a conversation by topic, check **Memories** or give it the exact title. See [Find conversations and control memories](https://bearly.ai/docs/features/memory-search.md) for the available settings.

## Create and organize chats

Ask for the result you want in ordinary language:

```text copy
Create a chat called Vendor review in the Operations project.
```

```text copy
Rename this chat to Q3 planning and move it to the Finance project.
```

```text copy
Move the chat called Interview notes out of its project.
```

Bearly creates a requested chat without opening it, so you stay in the conversation where you made the request. When you say “this chat,” Bearly applies the change to the current conversation.

If several chats have similar titles, Bearly may inspect the list or ask which one you mean before making a change.

## Delete chats

Ask from a different chat than the one you want to remove:

```text copy
Delete the old Vendor review and Pricing scratchpad chats.
```

Bearly shows a confirmation with the selected chat titles. Review the list and choose **Allow once** only when you want those exact conversations deleted.

> **Danger — Deleting chats is permanent**
> Deletion removes the selected conversations, their messages, chat-local files, and chat memory. It cannot be undone. Bearly cannot delete the chat that is currently executing the request; use another chat or the sidebar instead.


To organize conversations manually from the sidebar, see [Organize your chats](https://bearly.ai/docs/chats/organize.md).

For the product reasoning behind conversation search, source reading, and guarded chat actions, read [Making agents harness-aware, Part 1](/blog/making-agents-harness-aware-part-1).

---

---
title: "Search the web, find images, and run Deep Research"
description: "Find current information and visual references in chat, or plan a longer investigation that produces a sourced report."
canonical: "https://bearly.ai/docs/features/web-search-deep-research/"
markdown: "https://bearly.ai/docs/features/web-search-deep-research.md"
---

# Search the web, find images, and run Deep Research

HTML: [https://bearly.ai/docs/features/web-search-deep-research/](https://bearly.ai/docs/features/web-search-deep-research/)

> Find current information and visual references in chat, or plan a longer investigation that produces a sourced report.

Bearly offers three ways to work with public web information:

- **Web Search** is part of an ordinary chat reply. Bearly searches when current information or outside sources would improve the answer.
- **Image Search** adds a visual gallery when seeing examples would help, or searches for images on its own when the request is mainly visual.
- **Deep Research** is a separate, longer investigation. You review an editable research plan before starting a paid run, then receive a sourced report and PDF.

All three work in Bearly on the web, desktop, iOS, and Android when they are available to your account. An organization policy can set or lock Web Search and can disable Image Search or Deep Research.

## Get a current, sourced answer

With Web Search set to **Auto**, ask for the result you want in ordinary language. It helps to name the date range, geography, comparison, or kind of source that matters.

```text copy
Compare the three most recent quarterly results from these companies. Use their investor-relations releases and cite every figure.
```

Bearly decides when a search is useful, may open relevant pages for more context, and includes citations in its answer. Select a citation to open its source.

To control Web Search for the current chat:

1. Select the sliders control beside the chat input.
2. Under **Search**, select **Web**.
3. Choose **Auto** or **Off**.

**Auto** makes web tools available; it does not force every message to search. **Off** removes web search and page retrieval from the current chat, so answers may rely on the conversation and the model's existing knowledge instead.

To set the starting mode for future chats, open the sliders control, select **Chat Settings**, then choose **New Chats** → **Web Search** → **Auto** or **Off**. This default does not rewrite existing chats. Project chats may use their project's chat settings instead.

> **Note — When a setting is managed**
> If **Web Search** is missing from **This Chat**, or changing **Web** does not change the active mode, a project or organization policy may be setting it for you. A organization administrator can review the assigned policy under **Organizations** → organization name → **Policies**.


## Find visual references

Ask for images in ordinary language. With Web Search set to **Auto**, Bearly can pair a gallery with sourced web results or run an image-only search when the answer is mainly visual.

```text copy
Compare brutalist and mid-century modern desk lamps. Explain the distinguishing features, cite useful sources, and show visual examples of each.
```

Bearly leaves images out when they would not materially help. A general factual or news question should normally return sources without a decorative gallery.

Select an image to open a larger preview. From the preview, you can zoom, copy the image, download it, or open the page it came from. The source button on each thumbnail opens that page directly. Search images can disappear or change at their source; if a preview is unavailable, open the source or ask for another result.

To use a result in later work, refer to it by its position or source and state what you want done:

```text copy
Use the third image as a reference and create a warmer product-photo variation with a plain background.
```

Bearly first imports that result into the chat as an asset, then passes the asset to the image tool. During the initial search, Bearly selects results from their search metadata rather than inspecting every image's pixels. Ask it to import and analyze a result before relying on fine visual details.

Gallery images load through Bearly instead of connecting your device directly to the image search host. Opening a source is a normal visit to that website and is subject to the site's own privacy practices.

For a closer look at the gallery and how a selected result moves into the next task, read [Image search, in the same conversation](/blog/image-search-in-the-conversation).

## Run Deep Research

Use Deep Research for a question that deserves an explicit scope, a reviewable plan, and a substantial report—not for a quick fact check.



1. **Start with the Deep Research command**

   Type `/deep-research`, choose **Deep Research**, and enter the topic. On wider screens, you can instead select the **Deep Research** magnifying glass beside the chat input.

   ```text copy
   /deep-research Assess the US market for community solar software. Focus on 2024–2026 adoption, buyers, regulation, and credible market-size estimates.
   ```

2. **Review the research brief**

   Bearly prepares a **Deep Research** card with an editable **Query**. Clarify the time period, geography, definitions, exclusions, and comparison criteria before selecting **Create Plan**.

   Creating the plan does not use your allowance. An empty query cannot be submitted.

3. **Shape the plan**

   When **Plan** is ready, read it before starting the investigation. You can edit the plan directly. To ask for a newly drafted version, add guidance under **Notes**, then select **Revise Plan**.

   **Revise Plan** remains unavailable until you add notes. Direct edits in **Plan** are included in the eventual run even if you do not request another revision.

4. **Choose the depth**

   The card shows the selected tier and whether your plan includes it:

   - **Standard** requires the **Analyst** plan or higher.
   - Turn on **Ultra** for a deeper run. Ultra requires the **Research** plan.

   Each run uses your plan allowance based on its actual usage, up to a per-run cap.

   Planning and revision remain available before an upgrade. If the chosen tier is not included in your plan, **Run Research** opens the upgrade flow instead of starting the run.

5. **Start the investigation**

   Select **Run Research** only after the query, plan, and tier look right. The approved plan locks and the research begins.

   The research card shows its phase and elapsed time. You can open individual research-step chips while work is underway. A completed report hides those steps by default; select **Show research steps** to revisit them.

6. **Read and verify the report**

   The finished report appears in the chat with numbered citations and a collapsed source card. Select the source count to open **Sources**, then select any source to inspect it.

   Bearly also prepares a PDF. When delivery succeeds, the card shows **Sent to your inbox** and **View PDF**. If email delivery fails but the PDF was created, **View PDF** remains available in the chat.



> **Caution — Review before you run**
> The current Deep Research card does not offer a cancel button after a paid run begins. Treat **Run Research** as the final confirmation. The ordinary chat **Stop** control can stop a normal response, but do not rely on it to cancel a Deep Research run that has already started.


## Leave and return to a long run

Deep Research is designed to recover an in-progress run after a page refresh, app restart, or handoff to another signed-in device. Reopen the same chat to see the current status. Recovery can take a moment while Bearly reconnects to the investigation.

On the web, a long-running report may offer **Get the app**. The desktop app can keep the live status visible while you work elsewhere, but it is not required to receive the emailed report. On mobile, the operating system may pause the app in the background; reopen the same chat to resume its live status.

## If research does not go as expected

| What you see | What to do |
| --- | --- |
| Web Search returns no useful sources | Rephrase the request with distinctive names, a date range, location, or preferred source type. Ask Bearly to search again. |
| Image Search returns no useful images | Name the subject, visual style, viewpoint, location, or period more precisely. Ask for one new image search rather than repeatedly retrying the same query. |
| An image preview is unavailable | Open its source page or choose another result. Search images can be removed or replaced by their source. |
| A page cannot be read | Ask Bearly to use another primary source. Some sites block automated retrieval or require a sign-in. |
| **Deep Research** does not appear after typing `/deep-research` | Check whether your assigned policy permits Deep Research. If it should be available, reopen or update Bearly and try again. |
| **Create Plan** is unavailable | Add a non-empty question in **Query**. |
| Planning or revision reports an error | Keep the useful text, narrow the query or notes, and select **Create Plan** or **Revise Plan** again when the control is available. |
| **Run Research** opens billing | The selected tier is not included in the current plan. Use the indicated upgrade or choose Standard if your plan includes it. |
| The run shows **Research failed.** or **Research cancelled.** | The existing plan cannot be run again. Type `/deep-research` to begin a new investigation, reusing the useful scope from the earlier card. |
| The report email does not arrive | Return to the chat and select **View PDF** if it is available. Also check the inbox for the email address on your Bearly account. |

---

---
title: "Search academic papers with Scholar"
description: "Ask a research question and get an answer built on papers, with each claim cited to a passage you can open and check."
canonical: "https://bearly.ai/docs/features/scholar/"
markdown: "https://bearly.ai/docs/features/scholar.md"
---

# Search academic papers with Scholar

HTML: [https://bearly.ai/docs/features/scholar/](https://bearly.ai/docs/features/scholar/)

> Ask a research question and get an answer built on papers, with each claim cited to a passage you can open and check.

Scholar gives a chat a paper index beside Web Search. When a question should rest on research literature, Bearly searches papers, reads the passages that carry a claim, and cites them with superscript numbers. A **Papers** card under the reply lists what it found.

Scholar works in Bearly on the web, desktop, iOS, and Android, on every plan. A team's policy can turn it off.

## Ask a research question

Ask in ordinary language, the way you would put the question to a colleague in the field. Scholar is a better fit than Web Search for scientific, medical, and technical questions, and for anything that asks what the research says.

```text copy
What does the research say about whether sleep deprivation impairs working memory in adults?
```

Bearly decides when papers are the right source. During the reply you see **Searching papers**, then **Found 8 papers**, and often **Reading paper** as it pulls passages from the papers that carry the load-bearing claims. News, products, companies, and how-to questions still go to Web Search.

Citations appear as superscript numbers. Hover a number, or press and hold on a touch screen, to see the paper: where it lives, its title, authors, year, and the opening of its abstract. Select the number to open the paper.

## Read the Papers card

The **Papers** card under a reply names what was searched and how many papers came back. Expand it to see each paper with its source, authors, year, and field. Select a paper to open it.

The source tells you what kind of document you are looking at:

- **arXiv**, **bioRxiv**, and **medRxiv** are preprint servers. Bearly says so when a finding rests only on a preprint.
- **PubMed** and **PMC** are the biomedical literature.
- A website name, such as a journal's site, means Bearly found the paper through an academic web search rather than in the paper index. That search uses Web Search, so it needs Web Search available in the chat.

When Bearly reads a paper, the card marks it **Read** and quotes the passage it relied on. The step reads **Read 1 paper · 3 passages** when full text is available, or **Read 1 paper · abstract only** when the index holds only the paper's abstract, which is common for biomedical papers.

## Go deeper on one paper

Follow up in the same chat. Bearly can read a specific paper for a specific claim, or map the work around it.

```text copy
Give me the exact effect size from the first paper.
```

```text copy
Find later papers that cite or critique that paper.
```

Ask for follow-ups, critiques, or the prior work a paper builds on. To read a whole paper rather than passages, ask Bearly to read it; for open-access papers it can bring the full text into the chat as a document.

## Turn Scholar on or off

Scholar is on by default. To change it for the current chat:

1. Select the sliders control beside the chat input.
2. Under **Search**, select **Scholar**.
3. Choose **Auto** or **Off**.

**Auto** lets Bearly use papers when they fit the question. **Off** keeps papers out of the chat entirely. To use Scholar for one message while it is off, type `/scholar`, choose it from the list, add a space, and write your question.

> **Note — When Scholar is missing**
> If **Scholar** is not listed under **Search**, a team policy may have turned it off, or Web Search may be off for your team. A team administrator can review the assigned policy under **Teams** → team name → **Policies**.


## Limits

- Coverage comes from the paper index: arXiv, bioRxiv, medRxiv, and PubMed. Journal articles outside those sources can appear through the academic web search, listed by the journal's website, without passages.
- Full-text passages are available for many arXiv papers. Papers indexed only through PubMed give Bearly the abstract, so exact figures may need the paper opened.
- Bearly does not report citation counts or rank papers by prestige. It weighs study design, sample size, replication, and recency in the answer, and says when the literature is thin or inconclusive.
- **No papers found** means the index had nothing for that phrasing. Rephrase or broaden the question, or ask for the web instead.

Paper searches run through Bearly rather than connecting your device directly to the paper index. Opening a paper is a normal visit to that website.

---

---
title: "Watch and check YouTube videos"
description: "Paste a YouTube link into a chat to watch the video with a verdict, key moments, shorter cuts, and a summary, look up what it mentions, and check the figures it states."
canonical: "https://bearly.ai/docs/features/youtube/"
markdown: "https://bearly.ai/docs/features/youtube.md"
---

# Watch and check YouTube videos

HTML: [https://bearly.ai/docs/features/youtube/](https://bearly.ai/docs/features/youtube/)

> Paste a YouTube link into a chat to watch the video with a verdict, key moments, shorter cuts, and a summary, look up what it mentions, and check the figures it states.

Paste a YouTube link into a chat and Bearly opens the video in a card. The card has the player, a verdict on whether the video is worth your time, its key moments, shorter cuts, and tabs with the summary, takeaways, and what the video mentions. From the card you can look up the papers, books, repositories, and tools the video names, and check the figures it states against the web.

The card and its tabs work in Bearly on the web, desktop, iOS, and Android, on every plan. If a video can't play inside the card, select **Open on YouTube** to watch it there. Saving a video to play it and getting a missing transcript need the Bearly desktop app.

## Open a video

Paste the link into a chat, on its own or with a question. The card appears as soon as Bearly starts reading, so you can play the video while the tabs fill in. The reply answers your question. If you send only the link, the reply says in a sentence whether the video is worth watching. When Bearly can read the captions, it saves the transcript to the chat's files.

Links to videos, Shorts, and live streams work, from youtube.com or youtu.be. A link with a start time plays from that time. Playlists, channels, and clips aren't supported.

To make sure Bearly opens the video, type `/youtube`, select **YouTube** under **Tools**, and paste the link.

## Read the card

The header shows the title, channel, length, views, and upload month. Under it is the verdict, **Watch**, **Skim**, or **Skip**, with a sentence explaining it. When the video doesn't fully deliver what its title promises, **Title check** says so and gives a plainer title.

Under the player:

- When the video is long enough, **Watch** offers **2 min**, **5 min**, or **10 min** beside **Full video**. A shorter length plays the essential parts back to back in the same player, and the line beside it shows which part is playing.
- The playback speed menu and **Open on YouTube** sit at the end of the same row.
- The **Key moments** bar marks the parts worth watching. Select one to jump there. When YouTube has the data, a shaded curve shows which parts viewers replay most.

Select any time in the tabs to play from that moment. Only one video plays at a time in a chat, so starting another returns the first to its thumbnail. On a wide card, the expand button in the video's top corner fills the card with the video; select it again to go back.

The tabs depend on the video. A tab with nothing to show is left out.

| Tab | What it shows |
| --- | --- |
| **Summary** | The substance in a few sentences, who the video is best for and who can skip it, and the chapters. |
| **Insights** | Specific takeaways, each with the moment it's said and, when the captions were read, the words from the video. **Check claims** sits at the top when it's available. |
| **Claims** | The figures the video states and how many sites give the same figure or a different one. It appears when you select **Check claims** and stays with the chat. |
| **Mentioned** | The papers, books, code repositories, and tools the video names. |
| **Community** | What the comments agree on, push back on, correct, or add, and the moments people point to. Commenters aren't named. |
| **Transcript** | The full transcript. Search it, copy it, or select a line's time to play from there. |
| **Steps** | For how-to videos: the steps, commands you can copy, the tools used, and links. |

## Look up what a video mentions

**Mentioned** lists what the video names under **Papers**, **Books**, **Repositories**, and **Tools**, in the order they come up. Each row has the moment it's named and, when the captions were read, the sentence it's named in. Anything named only in a sponsor read or an ad is left out.

Select **Find** beside a row to look it up. Find shows the page it found and the site it's on, with the year for a paper. Select the result to open the page. Find only accepts a page that names the item, so when none does, it says **No match found.** rather than guess.

Find results aren't saved. After you reload the chat, **Find** appears again.

## Check the figures a video states

**Check claims** lists the figures the video states, such as statistics, amounts, counts, and dates, and searches the web for each one. It needs the video's transcript.



1. Open the **Insights** tab and select **Check claims**.

   The **Claims** tab opens and shows **Finding the figures in the video…**, then each figure as it's checked. Select **Cancel** to stop. Cancelling a first check saves nothing.

2. Read the result.

   The top line says how many figures were checked, followed by a count for each label. Each row has the moment the figure is said, the claim, and its label.

3. Select a row's label to see the sources.

   It shows the words from the video, then each site with the figure it gives, the words it uses, and a link to the page.

4. If the video states more than 12 figures, select **Check _N_ more** to check the next ones.



Labels count sites in the search results. Bearly doesn't decide whether a figure is true.

| Label | What it means |
| --- | --- |
| **Matched by _N_ sites** | Two or more sites give the same figure, and none gives a different one. |
| **Found on 1 site** | One site gives the same figure. Check it before you rely on it. |
| **Split · _N_ match, _N_ differ** | Some sites give the same figure and some give a different one. The check doesn't pick between them. |
| **_N_ sites give a different figure** | No result gives the video's figure, and at least one site gives a different one. |
| **Not found in _N_ results** | None of the results gives the figure. That doesn't make it wrong. When the search found nothing, or only pages about this video, the label says so. |
| **Couldn't check** | The check for that figure failed. **Check _N_ more** doesn't retry it. |
| **Not checked yet** | The figure is waiting for **Check _N_ more**. |

Each press checks up to 12 figures, and one check covers up to 40. The check is saved as a file in the chat's files, so it's still there after a reload and on your other devices. **Copy** at the top of the Claims tab copies the report with its sources. When the transcript came from automatic captions, the check says so, since automatic captions can mishear numbers. For a very long video, it says how far into the transcript it got.

## Ask about the video

Follow up in the same chat. Bearly answers from what it already read, without opening the video again.

- Ask what was said about a topic, and Bearly looks it up in the transcript.
- Ask for just one part, or a shorter version, and Bearly adds a card for that cut. Select **Play cut** to play its parts back to back, or select a time in its list to start there.
- After you check claims, ask about the results, such as which figures sites disagreed with. Bearly reads the saved check. It doesn't run a check itself; only you can, from the card.

## Copy the card

Select **Copy** under Bearly's reply. Along with the reply, it copies the card as text: the title and link, the verdict, the summary, chapters, insights with their quotes, the Mentioned list, community notes, and steps. Each time links to that moment on YouTube. The transcript and the claims check aren't included; use **Copy transcript** in the Transcript tab or **Copy** in the Claims tab for those.

## When the captions can't be read

YouTube doesn't always let Bearly read a video's captions. Bearly then watches the video to make the card, and **Insights** and **Mentioned** say so. Times are approximate, insights have no quotes, and names may be misheard. For a video longer than 90 minutes, the card covers the first 90 minutes.

On the web and on phones, a card made by watching has no **Transcript** tab or **Check claims** until a transcript is saved from the desktop app. If the video has captions that Bearly couldn't read, the desktop app can get them:



1. In the Bearly desktop app, open the card's **Transcript** tab. It shows **No transcript yet.**
2. Select **Get transcript**. The tab says what it's doing while the rest of the card keeps working. Select **Cancel** to stop.
3. The transcript appears in the tab and is saved to the chat's files.



Once it's saved, the transcript shows on the card on your other devices, **Check claims** becomes available, and follow-up questions can quote it. The summary and insights keep their approximate times. If the video has no captions at all, the tab says **This video has no captions.**

**Get transcript** isn't offered in a chat you can only view, for a live stream, or for a video YouTube reports as having no captions. Otherwise, if a card made by watching has no **Transcript** tab in the desktop app, update the app.

## When a video can't play in the card

Some uploaders don't allow their videos to play outside YouTube. The card then says **This video can't play in an embedded player.** and offers **Open on YouTube**. Outside the desktop app, selecting a time on the card opens the video on YouTube at that moment.

In the desktop app, Bearly can save a copy of such a video to the chat and play it from there. Select **Download to play here**. The player shows the progress, and **Cancel** stops the download. Bearly never starts a download on its own.

Many videos can't be saved this way. When a save fails, the card says **The download couldn't start.** or **The download didn't finish.** and still offers **Open on YouTube**.

The copy is saved to the chat's files and plays on your other devices too. Bearly saves a lower quality when needed to stay under the 300 MB limit for a chat file. If even that is too large, the card shows **This video is over the 300 MB limit for a chat file.** To save videos as files on purpose, see [Download a video](https://bearly.ai/docs/features/download-video.md).

## Plans, allowance, and team policy

- **Plans:** YouTube cards are available on every plan. Opening a video uses your plan allowance, like any other reply.
- **Allowance:** **Find** and **Check claims** also use your plan allowance, and **Check claims** says **Uses plan allowance.** beside the button. When your allowance is used up, **Check claims** isn't offered and **Mentioned** says **Your plan allowance is used up, so Find is off for now.** A new video whose captions Bearly can't read can't be summarized then, and the card says **The summary couldn't be made for this video.** See [Manage billing and review usage](https://bearly.ai/docs/account/billing-usage.md).
- **Chats you can only view:** In a chat you can view but not edit, the card plays and its tabs can be read, but **Find**, **Check claims**, **Get transcript**, and saving a video aren't available.
- **File uploads:** The transcript, a saved video, and a claims check are files in the chat. When a managed organization's policy turns off **File uploads** (under **Privacy & Security**), Bearly saves none of them, so the card has no **Transcript** tab, **Get transcript**, **Check claims**, or **Download to play here**. The rest of the card still works, and files saved earlier still open.
- **Routines:** Scheduled [Routines](https://bearly.ai/docs/features/routines.md) don't open YouTube videos.
- **Team policy:** There's no chat setting for YouTube. Administrators of managed organizations control it from **Organizations** → organization name → **Policies** → policy name → **Web & Research** → **YouTube**, which needs **Web** on. **Find** and **Check claims** also need **Chat web search**. Without it, **Find** can still look up papers when [Scholar](https://bearly.ai/docs/features/scholar.md) is on.

Turning **YouTube** off stops new links from opening as cards. Cards already in chats still show, without **Find**, **Check claims**, **Get transcript**, or saving a video.

## If something doesn't work

| What you see | What to do |
| --- | --- |
| **That doesn't look like a YouTube video link.** | Paste the link to a single video, not a playlist, channel, or clip. |
| **This video couldn't be opened. It may be private, removed or restricted.** | Open the link on YouTube to check that it still plays for you. Bearly can't read private or restricted videos. If it plays on YouTube, ask Bearly to open the video again. |
| **Reading was stopped before this was ready.** | The reply was stopped, or the page reloaded, before the card was finished. Ask Bearly to open the video again. |
| **The summary couldn't be made for this video.** | Ask Bearly to open the video again. If it fails again, check whether your plan allowance is used up in [Manage billing and review usage](https://bearly.ai/docs/account/billing-usage.md). |
| **This video can't play in this app.** or **This video can't be played here.** | Select **Open on YouTube** to watch it there. In the desktop app, you can instead select **Download to play here** when it's offered. |
| **The download couldn't start.** or **The download didn't finish.** | Select **Try again**. Many videos can't be saved to a chat; if it keeps failing, select **Open on YouTube**. |
| **Update the desktop app to download videos.** | Update the Bearly desktop app. |
| **The desktop app is saving other videos. Try again when one finishes.** | Wait for a download in another card to finish, then select **Try again**. |
| **Couldn't search right now.** under a **Mentioned** row | Select **Try again**. |
| **Find** doesn't appear in **Mentioned** | Find needs plan allowance and a chat you can edit. On a managed account, books, repositories, and tools need **Chat web search**; papers need **Chat web search** or **Scholar**. |
| **Couldn't check the claims.** | Select **Try again**. |
| **Checking claims needs plan allowance, and yours is used up.** | Figures checked before the allowance ran out are kept. When you have allowance again, reload the chat to continue. |
| **Checking claims isn't available on this account right now.** | Your organization's policy may not allow what the check needs. Ask an administrator. |
| **This check couldn't be saved to the chat, so it will be gone after a reload.** | Select **Copy** in the Claims tab to keep the report. |
| **Check claims** doesn't appear in **Insights** | The video needs a transcript, you need plan allowance, and the chat must be one you can edit. On a managed account, **Chat web search** and **File uploads** must be allowed. For a video Bearly watched, get the transcript in the desktop app first. |

---

---
title: "Look up a person with People Search"
description: "Confirm which person you mean, then get a cited brief of their public work with a source and date on every claim."
canonical: "https://bearly.ai/docs/features/people-search/"
markdown: "https://bearly.ai/docs/features/people-search.md"
---

# Look up a person with People Search

HTML: [https://bearly.ai/docs/features/people-search/](https://bearly.ai/docs/features/people-search/)

> Confirm which person you mean, then get a cited brief of their public work with a source and date on every claim.

> **Note — Analyst plan**
> People Search needs the **Analyst** plan or higher, and an organization policy can turn it off. If typing `/people-search` offers nothing, it is not enabled for your account.


People Search identifies one named person and builds a brief from public sources: their current role, career history, published work, public accounts, and — when your organization allows it — a way to reach them. Every claim carries the sources behind it and a date.

It runs in an ordinary chat on the web, desktop, iOS, and Android. On a lower plan the upgrade prompt appears in the chat instead of a search. It cannot run in a chat you are viewing read-only.

The cited brief works on one person at a time. For a list you already have, the separate **Lead enrichment** action can return structured professional profile data for up to ten leads at once.

## Enrich up to ten known leads

Use Lead enrichment when you already know who or what role you want and need consistent profile fields across a short list. A batch can mix any of these inputs:

- a public LinkedIn or X profile link, or a Facebook vanity profile link such as `facebook.com/jane.doe`
- a work or personal email address
- a first name and employer name or domain, with an optional last name, title, and location
- a role and company, such as `CTO at Acme`

Type `/lead-enrichment`, choose **Lead enrichment**, and paste or describe the rows. Bearly accepts one to ten leads in each batch. It refuses an eleventh row and does not split or continue a larger list on its own.

Before the lookup starts, review the number of leads and the maximum cost estimate, then select **Allow**. Successful rows return the public professional profile link, name, headline, location, summary, work history, and education that were available. Failed rows stay in the result with a plain reason, so one miss does not discard the rest of the batch.

Lead enrichment is not the cited brief: it does not read wider web sources, provide citations, or return personal email addresses or phone numbers. Use `/people-search` when you want Bearly to identify and confirm one person before building sourced research. It also does not search a broad prospect database from open-ended filters; each row must identify a person or one role at one company.

## Find the right person



1. **Start the search**

   Type `/people-search`, choose **People Search**, and give the name plus at least one detail that separates this person from everyone else with the same name: a company, a city, a school, a role, or a profile link. Asking for it in ordinary language works too.

   ```text copy
   /people-search Sarah Chen, product lead at Plaid in San Francisco
   ```

   Given only a name, Bearly asks for one detail before it searches. Given a profile link, an email address, or a handle, it may resolve to a single person and take you straight to the confirmation in step 4.

2. **Read the candidates**

   A **People Search** card appears with up to five people. **Looking for** restates what Bearly understood as chips, so you can see it read your request correctly. **Checked _n_ sources** counts the public sources read so far.

   Each row carries a photo, a headline and location, distinguishing facts, a one-line reason the person was included, and a result for each detail you gave, such as **MATCHES ROLE** or **NO DATA: SCHOOL**. Match strength is a word — **Strong match**, **Likely match**, or **Possible match** — never a percentage. If you searched on a name with no other detail, no row reads stronger than **Possible match** and the card says why.

   **Compare** puts candidates side by side. Columns are numbered in the card's order so you can tell namesakes apart, and they wrap to fit the space. Long values clamp, with the full text on hover. A column with nothing to show says only that nothing beyond the name came back. Comparing is free and does not spend a search.

3. **Narrow the list if none of them are right**

   **Add a detail** opens a panel: type the detail, check **Kind**, and select **Add detail**. **None of these** clears the list and searches again.

   Both re-run the search, and both spend one of the card's searches. The footer counter — **_n_ of _n_ searches used** — is the whole budget for this card. When it is spent, the card explains that it will not run again; pick someone from the list, or start a new search with a profile link, email, or handle in it, which narrows far better than another detail would.

   **Cancel** stops a search that is still running. Searches already spent stay counted.

4. **Confirm the person**

   Select **This is them**. A panel opens naming that person and the estimated cost of the lookups; **Build brief** starts them.

   This step is what the card is for. Building a brief is the expensive part of People Search, and it researches exactly one person — the one you confirm. Nothing is looked up about a candidate you passed over.



> **Caution — Read the reason line before you confirm**
> Public profile data contains merged and mislabeled records: one entry can carry one person's name over another person's profile link. When a candidate's name disagrees with its own link, the card says so in that row's reason, drops its strength, and will not resolve to that person on its own. Read the reason and open the profile link before confirming.


The card then collapses to a single confirmed row, and the brief assembles below it.

## What a brief contains

While it builds, a row of chips shows each lookup — **Web presence**, **Writing & talks**, **X history**, **GitHub**, **Contact route** — with its state and a plain sentence about what it found, skipped, or could not finish. Mid-run, empty tiles read **Gathering…** and unwritten sections say what is still coming; chips like **Corroborated** appear only after the run finishes. One lookup coming up empty does not stop the rest, and once the brief is up, whatever was blocked or did not finish is folded into one plain sentence near its top — a lookup that simply found nothing is an answer, not a failure, and is not listed there.

The finished brief opens with a photo, name, headline, location, an **Updated** date, and **Refresh**, over four tabs:

- **Overview** — four fast facts (**Current role**, **Employer**, **Location**, **Last active**), each with its source and date; a short written summary with numbered citations; and a **Career** timeline of dated roles and schooling.
- **Work & ideas** — a cited account of the person's work, **Recurring themes**, and the dated record grouped under **Shipped**, **Talks & panels**, and **Writing**, each entry with its venue and date; anything else lands under **Other work** or **Other dated work**. Long lists collapse behind **See _n_ more**.
- **Public presence** — an **Accounts** card per platform with the dates that platform publishes and a follower range rather than a follower count, a dated **Activity** list, and a section headed **From their public writing**: how the person writes in public, not a conclusion about who they are.
- **Sources & confidence** — every claim beside its sources. A chip row grades how well each area is covered, from the same claims listed under it. Every claim is a row; expand it for the verbatim quote behind each source. The full source list sits below, each entry with its domain and date.

Citation numbers throughout the brief jump to **Sources & confidence** and highlight the matching source row; the row itself opens the page.

**Reach out** sits under the sections when your organization allows contact routes. The best route is on top, with the value masked until you select **Reveal** and a chip saying in plain language what was and was not checked — **Accepts mail** means a verifier confirmed the inbox accepts mail, and nothing more. Revealing is free; the value is already in the brief. Alternates and an **Also on** row of public profiles follow.

Bearly never guesses an X handle or a GitHub login from a name. If no account was found on the record, that lookup is skipped and says so.

## Every claim carries its source and date

Facts are dated individually, not by when the run happened. A tile reads **as of** the month its source published when the page carried a date, and the date Bearly read the page when it did not. A fact whose newest source is old is tagged **past**.

In **Sources & confidence**, each claim carries one of four labels:

| Label | What it means |
| --- | --- |
| **Corroborated** | Two or more independent sources state the same thing. |
| **Partial · inferred** | One source only, or a value inferred rather than read directly. Check it before you rely on it. |
| **No data** | Nothing was found either way. This is not a failed lookup. |
| **Conflicting** | Sources disagree. Both values are kept. |

## Contradictions are shown, not resolved

When sources disagree about a fact, the Overview tile shows a **Conflicting** chip that takes you to **Sources & confidence**, where both values stay visible with their own dates and sources and a line saying the brief does not choose between them.

Bearly does not choose between them, average them, or drop the older one. A person who changed jobs last month and a directory that has not caught up are both telling the truth about different moments, and which one matters depends on why you are asking. The written summary says the fact is unclear rather than asserting one side.

## Refresh a brief

**Refresh** re-runs every lookup for the person you confirmed and reports what changed. It costs what the first build cost, so it asks again, with the estimate shown, before it starts.

The previous brief is kept, not overwritten. When the new run finishes, a **What changed** strip appears above the tabs with up to ten lines marked **New**, **Changed**, or **No longer reported**, each with its own citations. Closing the strip keeps the brief.

Refresh before outreach, or after news you expect to have changed the person's role. There is no reason to refresh on a schedule — nothing in a brief changes hour to hour, and each refresh is billed.

## Check public records with On the record

**On the record** is a collapsed band at the bottom of the brief, and a separate paid action from the brief itself. It runs one check for US public records about the person you confirmed — criminal and corrections records, plus any court filings or other findings that come back with them. It never runs as part of a build or a refresh, and nothing it finds is written into the brief's summary.



1. **Expand the band**

   It states what the check covers, offers an optional **Date of birth** field, and carries the informational-use notice below.

2. **Add a date of birth if you have one**

   Optional, and used for this check only. It narrows the match, and it is the strongest single detail you can give: a record that matches on name and date of birth is a different thing from one that matches on name alone.

3. **Run the check**

   Select **Run records check**, then **Confirm** on the estimate the band shows. The check needs a first and last name plus a city, a US state, or a date of birth; when it does not have enough to search against, the band says so instead of running.



Findings come back grouped by the kind of record they are, each with a jurisdiction, a date, and a plain sentence about what was matched:

- **Strong match** — the name plus a second identifier lines up with the person you confirmed. When some records tie and others do not, the collapsed band counts both: **_n_ records · _m_ unverified**.
- **Possible · unverified** — kept in its own muted group, **Possible matches — unverified**. These share a name and nothing else that could be checked; any of them may belong to a different person with the same name. Rows with nothing to show beyond their kind fold into one counted line each rather than stacking.

When not one returned record could be tied to the person you confirmed, the results say so up front, the collapsed band reads **None tied to this person · _n_ possible**, and the band offers **Check again** with its own **Date of birth** field — a date of birth can show whether those rows belong to this person. Checking again is a second paid check with its own confirmation.

A check that finds nothing is still a complete answer: the band reads **No public records found · checked _date_**. It is billed the same as a check that returns records.

> **Caution — Informational use only**
> The notice on this band is a limit on what you may do with the result, not boilerplate: **For personal, informational use only — not for decisions about employment, housing, credit, or insurance (FCRA).** These records are not a background check and are not assembled to the standard the law requires for those decisions. If you are hiring, renting, lending, or underwriting, use a consumer reporting agency instead.
>
> These sources cover US jurisdictions only. Records filed elsewhere, or under a different name, do not appear.


## What People Search cannot do

- **It works best on people with a public professional presence.** Someone who publishes, speaks, ships code, or holds a role their employer names in public is well covered. Someone whose professional life appears only in a licensing registry, or barely appears at all, may come back thin, or not at all.
- **It can fail to find the right person.** Common names, people who share a name with someone more prominent, and people who keep a small public presence are the hard cases. When nothing matches, the card says so rather than filling the space; add a detail or start again with a profile link.
- **A brief is a reading of public sources, not a verified record.** Read the reason on the card and the sources in **Sources & confidence** before acting on anything consequential. A **Partial · inferred** claim is one page's word.
- **It is not for screening people.** People Search is built to help you recognize and prepare for someone — a candidate you are about to interview, an author you are about to email, a founder you are about to meet. It is not a background check, and the records band's limits are not a formality.
- **A shared chat opened read-only can be read, not run.** Open your own chat to start a search, refresh a brief, or run a records check.

## If a search does not go as expected

| What you see | What to do |
| --- | --- |
| `/people-search` offers nothing | Your assigned policy turns People Search off. If that policy disables features by default, an administrator has to switch it on. |
| Bearly says People Search needs the Analyst plan | The upgrade prompt is already in the chat. People Search needs **Analyst** or higher. |
| Bearly asks for a detail instead of searching | Add a company, city, school, role, or profile link. On a bare name the result would be a guess. |
| **Nobody came back for this name and these details** | Add a detail, or search again with a profile link, email address, or handle. |
| Every candidate reads **Possible match** | The search had no detail to work with, or nothing corroborated one. Add one rather than trusting the top row. |
| The card says its searches are used up | It will not search again. Pick someone from the list, or start a new search — with a link, email, or handle if you have one. |
| A lookup chip reads skipped or did not finish | Select the chip for the reason. A missing account, a source that did not answer, or an organization policy each read differently. Refresh re-runs them. |
| **Reach out** is missing entirely | Your assigned policy turns off contact routes. |
| **On the record** is missing entirely | Your assigned policy turns off public records. |
| The records band will not run | It needs a first and last name plus a city, a US state, or a date of birth. Add a date of birth if the location cannot be placed in the US. |
| A records check is still **Checking public records…** long after you closed the tab | Reopen the chat and give it a few minutes. A check whose tab went away is released so the band can run it again. |
| The findings from a completed check are gone | The band says when the findings are no longer stored in the chat. Run the check again if you still need them. |

Administrators of managed organizations control availability from **Organizations** → organization name → **Policies** → policy name, under **Web & Research**: a **People Search** switch, plus separate **People Search contact routes** and **People Search public records** switches beneath it. Turning **People Search** off hides the other two. A switch nobody has set follows the policy's own default — **Enabled by default** or **Disabled by default**.

---

---
title: "Research your documents"
description: "Ask focused questions across attached or project files, compare sources, and open cited passages for verification."
canonical: "https://bearly.ai/docs/features/document-research/"
markdown: "https://bearly.ai/docs/features/document-research.md"
---

# Research your documents

HTML: [https://bearly.ai/docs/features/document-research/](https://bearly.ai/docs/features/document-research/)

> Ask focused questions across attached or project files, compare sources, and open cited passages for verification.

Bearly can find and combine relevant passages from documents in a chat. Ask a focused question and the answer can include numbered citations that lead back to the source file and page.

Use this for evidence gathering, comparison, and precise questions. If you want a broad overview of one long file, [create a Deep Summary](https://bearly.ai/docs/features/deep-summary.md) instead.

## Ask about attached documents



1. **Attach the source material**

   Select the plus icon beside the message box, choose **Attach files**, and add the documents. Wait for each attachment to finish loading.

2. **Ask a focused question**

   Name the files, evidence, and form of answer that matter. Bearly can work across several sources in one request.

   ```text copy
   Compare the renewal and termination terms in these agreements. Make a table with notice periods, fees, and unusual exceptions, and cite every row.
   ```

3. **Send the message**

   Bearly reads or searches the relevant material and answers in the chat. Preparing a newly attached or scanned document can take a moment.

4. **Open the evidence**

   Select a numbered citation in the answer—or a source filename shown above it—to open the document. When a page is known, Bearly opens near that page so you can inspect the surrounding text.



For a follow-up, stay in the same chat. You can ask Bearly to quote a passage, explain a term, resolve a disagreement between files, extract a table, or look for evidence that would challenge the first answer.

## Research project files

Documents added to a project are available to chats in that project. Open or start a project chat and refer to the material by name:

```text copy
Using the documents in this project, trace how the proposed launch date changed and identify who approved each revision. Cite the source for every date.
```

This is useful for a continuing body of reference material because you do not have to attach the same documents to every project chat. Files attached only to one ordinary chat remain with that chat.

## Work with strong source material

- Use descriptive filenames so citations are easy to recognize.
- Keep only relevant sources in scope, or name the exact files to compare.
- Searchable PDFs, word-processing documents, presentations, spreadsheets, CSV, Markdown, and plain-text files are common choices. A document can be up to 200 MB for document search.
- Scans and unusual formats may need extra processing. If text cannot be extracted, export a searchable PDF or paste the important passage into the chat.
- For audio or video, use [Transcribe](https://bearly.ai/docs/features/transcripts.md) before asking detailed questions about the spoken content.

> **Caution**
> Citations make an answer easier to verify; they do not guarantee that every inference is correct. Open the cited passage before relying on an exact quotation, number, deadline, legal term, or safety-critical claim.


## If Bearly cannot find the answer

- Check that the upload completed and that you are in the chat or project containing the file.
- Mention the document name and use distinctive words, dates, people, or section titles from the likely passage.
- Ask a narrower question first, then request a synthesis after the relevant evidence has been found.
- If a citation no longer opens, the source file may have been removed. Attach it again and repeat the question.
- If no passage is found in a scan, attach a text-searchable copy.

For help attaching, previewing, or downloading a source, see [Work with files in a chat](https://bearly.ai/docs/chats/files.md). For reusable project material, see [Add documents to a project](https://bearly.ai/docs/projects/documents.md).

---

---
title: "Summarize a document"
description: "Create a structured Deep Summary of a long document, then continue working with it in the same chat."
canonical: "https://bearly.ai/docs/features/deep-summary/"
markdown: "https://bearly.ai/docs/features/deep-summary.md"
---

# Summarize a document

HTML: [https://bearly.ai/docs/features/deep-summary/](https://bearly.ai/docs/features/deep-summary/)

> Create a structured Deep Summary of a long document, then continue working with it in the same chat.

**Deep Summary** condenses a document into a readable overview. It is useful when you want the main argument, important sections, decisions, or themes before examining individual details.

For a precise question, quotation, or source comparison, ask Bearly to read or search the document instead of summarizing it.

## Create a Deep Summary



1. **Attach the document**

   Open a chat, select the plus icon beside the message box, choose **Attach files**, and select the document. Wait until the file finishes attaching.

2. **Choose Deep Summary**

   Type `/summarize`, then select **Deep Summary** from the command menu.

3. **Describe the summary you need**

   You can request a general overview or identify what should receive the most attention.

   ```text copy
   Summarize this report for an executive reader. Preserve the central argument, major evidence, risks, and every explicit recommendation.
   ```

4. **Send the message**

   Bearly reads the document and writes the summary into the conversation. Long documents can take more time because their sections are processed before the final summary is complete.



Deep Summary is clearest when you summarize one document at a time. If several files are attached, name the file you want summarized.

## Continue from the summary

The summary remains part of the chat, so you can ask Bearly to:

- Expand one section or explain an unfamiliar term.
- Turn the summary into a briefing, outline, email, or list of questions.
- Compare a claim against the attached source.
- Identify details that the condensed summary may have omitted.

> **Caution**
> A summary is a condensed interpretation, not a substitute for reviewing the source when exact wording or complete detail matters. Ask Bearly to read the relevant passage before relying on a quotation, number, deadline, or legal obligation.


## If Bearly cannot summarize the file

- Confirm that the upload finished and send the request again.
- If the document is password-protected, unlock it and attach an accessible copy.
- If the file contains scanned pages, allow more time for the text to become available.
- If Bearly says it cannot extract the document content, copy the relevant text into the chat or export the document to PDF and try again.
- For a very large file, ask for the most important sections first, then request a broader synthesis.

Learn how Bearly works with source material in [Document Research](https://bearly.ai/docs/features/document-research.md).

---

---
title: "Find conversations and control memories"
description: "Search your Bearly chat history and choose which earlier conversations a chat can use for context."
canonical: "https://bearly.ai/docs/features/memory-search/"
markdown: "https://bearly.ai/docs/features/memory-search.md"
---

# Find conversations and control memories

HTML: [https://bearly.ai/docs/features/memory-search/](https://bearly.ai/docs/features/memory-search/)

> Search your Bearly chat history and choose which earlier conversations a chat can use for context.

Bearly offers two related controls: **Search conversations** helps you find and reopen an earlier chat, while **Memories** lets the AI look for relevant context from earlier conversations as it answers.

Changing **Memories** does not change which conversations appear in the search window.

## Find an earlier conversation



1. **Open conversation search**

   Select the **Search conversations** magnifying glass in the chat sidebar. You can also press <kbd>⌘</kbd> + <kbd>F</kbd> on a Mac or <kbd>Ctrl</kbd> + <kbd>F</kbd> on Windows and Linux.

2. **Describe what you remember**

   Enter a phrase in **Search conversations...**. You can search for a topic or idea even when you do not remember the exact wording.

3. **Open a result**

   Results are grouped by conversation. Each result shows the chat title, its number of matches, and a preview of the most relevant matching message. Select a result to open that chat.



Select **Clear search** to remove the current phrase, or press <kbd>Esc</kbd> to close search. If Bearly shows **No results found**, try a broader description or a distinctive phrase from the conversation.

## Choose what a chat can remember

In a chat, select the sliders icon beside the message box, then open **Memories**.

- **Off** prevents the chat from searching earlier conversations for context.
- **Project Only** lets a project chat use memories from chats in the same project.
- **All Chats** lets the chat search across your Bearly conversations.

Outside a project, the available choices are **Off** and **All Chats** because there is no project boundary to use.

When memories are available, Bearly may search them when your request refers to earlier work, decisions, preferences, or missing context. You can also ask directly:

```text copy
Find the budget assumptions we agreed on in an earlier conversation and use them in this forecast.
```

> **Note**
> The first search across older chats can take longer while Bearly prepares existing conversations. New conversation content becomes searchable automatically.


## Understand defaults and policy locks

The selection applies to the current chat. Defaults for new chats can be changed in **Chat Settings** → **New Chats**.

A project setting or managed-team policy may supply or lock the memory behavior. If **Memories** will not change, review the project's settings or ask a team administrator which policy applies.

## If a memory is missing or unwanted

- If Bearly cannot find an earlier detail, use **Search conversations** to locate the original chat and continue there, or paste the relevant information into the current message.
- If the recalled context is not relevant, select **Memories** → **Off** and restate the request without relying on earlier conversations.
- If search has no results immediately after opening Bearly on a new device, leave the app open briefly and try again after earlier chats finish preparing.

---

---
title: "Work with local files"
description: "Let Bearly read, create, edit, and run files on your computer from a folder you choose."
canonical: "https://bearly.ai/docs/features/work-locally/"
markdown: "https://bearly.ai/docs/features/work-locally.md"
---

# Work with local files

HTML: [https://bearly.ai/docs/features/work-locally/](https://bearly.ai/docs/features/work-locally/)

> Let Bearly read, create, edit, and run files on your computer from a folder you choose.

**Work locally** lets Bearly work with files and commands on your computer while you remain in control of each kind of access. It is available in the Bearly desktop app for macOS, Windows, and Linux.

On the web, **Work locally** opens the desktop-app handoff. The control is not shown on mobile. Chats that used local files remain readable on other devices, but Bearly cannot inspect, refresh, or change those files until you return to the same chat in the desktop app.

> **Caution — Local commands are not confined to the selected folder**
> The selected folder gives Bearly a working location; it is not a security sandbox. A command you approve can reach other files that your operating-system account can access. Review the command, path, reason, and overwrite details before allowing an action.


## Complete a task in a local folder



1. **Open the chat in Bearly desktop**

   Start a new chat or return to the chat that owns the work. The **Work locally** control appears in the context strip attached to the chat input after the chat finishes loading.

2. **Choose where Bearly should work**

   Select **Work locally**, then choose **Choose folder…** and select the folder for the task. Bearly shows the folder name on the control. Previously selected folders may appear in the same menu, with their full paths, so you can return to one without opening the system picker.

   If you do not choose a folder, **Auto** can use a private work area for this chat. Choose a folder when the task should read or change an existing local project, or when you want the finished files placed somewhere specific.

3. **Describe the result and boundaries**

   Name the relevant files, the desired result, and anything Bearly must not change. Relative filenames refer to the current working folder.

   ```text copy
   Read the CSV files in this folder, create a monthly summary workbook, and add the finished workbook to this chat. Do not change the source files.
   ```

4. **Review each permission request**

   In Standard, Bearly pauses before reading, writing, editing, saving a chat file locally, or running a local command unless an applicable permission is already active. Adding a local file from outside the working folder also requires review. The card names the action and explains why it is needed. Select **Details** to inspect the file, command, working folder, overwrite choice, or other relevant information. [Full access](https://bearly.ai/docs/settings/agent-permissions.md) skips these application approvals for enabled local tools; it does not restrict commands to the selected folder.

   - **Allow** approves only the action in front of you.
   - **Deny** blocks that action and lets Bearly continue without it when possible.
   - **Allow for…** lets you choose **This response**, **For now**, or **For this chat**. Read the expanded warning, then use the confirmation button to grant that duration.

   Read-only grants use labels such as **Allow reads for this response**. A broader local-tools grant can also cover writes, edits, file transfers, and commands. **Auto** makes local work available to Bearly; it does not approve these requests for you.

5. **Inspect the result**

   Bearly reports what it read, changed, created, or ran. Ask it to verify important output—for example, reopen a generated file, summarize changes, or run a relevant check.

   For a finished file you want to keep with the conversation, ask Bearly to “add it to this chat.” The file then appears as a normal chat file you can open or download later.

6. **Stop or revoke access**

   Select the chat's **Stop** control to stop the current response. If a local command is running, Bearly also stops that command.

   When a session- or chat-length grant is active, the context strip shows **Local reads allowed** or **Local tools allowed**. Select its **Revoke** control to remove the active session and chat permissions. Bearly will ask again before another protected action.



## Change folders or turn local work off

Select the current folder name or **Work locally** to reopen its menu.

- Choose another recent folder to make it the working folder for this chat.
- Select **Choose folder…** to browse somewhere else.
- Select **Clear folder** to return to the chat's private work area.
- Select **Off for this chat** to remove local file and command access from Bearly's available tools.
- When it is off, select **Turn on for this chat** to restore **Auto** mode.

The selected folder and chat-length permissions belong to this chat on this device. Opening a different chat does not silently give it the same working folder or permissions.

## Move files between the computer and chat

Work locally can bridge the two places when you ask clearly:

- “Add `final-report.pdf` to this chat” publishes a local result as a chat file. A file outside the working folder requires review before its contents are uploaded.
- “Save the attached workbook as `inputs/source.xlsx` in this folder” copies a chat file into the working folder after you approve the write.
- “Update the local draft, then add the revised copy to the chat” changes the local file first and publishes the result separately.

Local file contents you approve for reading may be sent to the AI model working on the reply. A local file you add to the chat is stored with that conversation like other chat files.

## Work safely

- Use a narrowly scoped folder that contains only what the task needs.
- Ask Bearly to stop before overwriting, deleting, installing, sending, or publishing anything consequential.
- Check **Overwrite** in permission details. A write can replace an existing file.
- Prefer **Allow** for one action until you understand the flow. Use **This response**, **For now**, or **For this chat** only when repeated prompts would obstruct work you have already reviewed.
- Revoke a broad grant as soon as the task is finished.

For work in visible applications rather than files and commands, see [Use your computer with Bearly](https://bearly.ai/docs/features/computer-use.md).

## If local work is unavailable or fails

| What you see | What to do |
| --- | --- |
| **Work locally** opens **Get the app** | Install or open the Bearly desktop app, then return to the same chat. |
| The control does not appear in Bearly desktop | Update Bearly, reopen the app, and let the chat finish loading. Older desktop builds may not support all local actions. |
| A recent folder no longer works | The folder may have moved, been renamed, or been removed. Select **Choose folder…** and choose its current location. |
| Bearly cannot read or write a protected folder | Choose a folder your account can access, or grant Bearly the relevant operating-system file permission and retry. |
| You selected **Deny** by mistake | Ask Bearly to try the action again, review the new card, and select **Allow**. |
| A command fails or times out | Open **Details**, check the command and working folder, then ask Bearly to retry with a narrower step. Confirm that any program the task needs is available on your computer. |
| A file is too large to read directly | Ask Bearly to inspect only the relevant rows, pages, or text range, or work from a smaller copy. |
| Local work is blocked for your organization | Ask an administrator to review **Organizations** → organization name → **Policies** → policy name → **Code & Automation** → **Work locally**. |

---

---
title: "Use your computer with Bearly"
description: "Let Bearly complete tasks in visible desktop apps while you control what it can see, do, and how long access lasts."
canonical: "https://bearly.ai/docs/features/computer-use/"
markdown: "https://bearly.ai/docs/features/computer-use.md"
---

# Use your computer with Bearly

HTML: [https://bearly.ai/docs/features/computer-use/](https://bearly.ai/docs/features/computer-use/)

> Let Bearly complete tasks in visible desktop apps while you control what it can see, do, and how long access lasts.

Computer Use lets Bearly work through the same windows and controls you use: it can inspect what is on screen, click, type, scroll, drag, use keyboard shortcuts, and move between visible apps.

It is available in the Bearly desktop app for macOS, Windows, and Linux. It is not available on the web or on mobile. A conversation that used Computer Use remains readable elsewhere, but return to that chat in the desktop app before asking Bearly to look at or control your computer again.

For work that happens entirely in a browser, [Control Chrome from Bearly](https://bearly.ai/docs/features/chrome-control.md) is usually faster and more reliable than driving Chrome through the screen.

> **Caution — Computer Use can control the full desktop**
> Bearly may capture and send screenshots of your full display to the AI model selected for the task. It may also change focus, switch between visible apps, and control the mouse and keyboard. Close or hide anything you do not want included before granting access.


## Complete a desktop task



1. **Open a chat in the desktop app**

   For the clearest result, open the app or window involved and leave it visible. You do not need to turn on a separate mode.

2. **Describe the result and its boundaries**

   Ask for the complete task in ordinary language. Include the app, the desired result, and any point where Bearly should pause.

   ```text copy
   In the spreadsheet already open on my desktop, sort the Budget tab by due date. Stop before saving.
   ```

3. **Review the permission request**

   In Standard, when Bearly needs desktop access, a **Use your computer** card shows the task and requested desktop access. [Full access](https://bearly.ai/docs/settings/agent-permissions.md) skips eligible application approvals, but operating-system permissions and required safety checks still apply. When a card appears, check its details before choosing:

   - Choose **Once**, **This response**, **This session**, or **This chat**, then select the Allow button. Session access ends when the chat reloads or the app restarts. Some settings or account-policy changes also reload the chat. Chat access covers future responses in this chat on the same device until revoked.
   - A separate confirmation may still appear for a safety check raised by the Computer Use provider.
   - **Deny** keeps desktop access blocked. You can revise the request or complete the step yourself.

   Desktop access can include switching between visible windows and the full desktop while completing the approved task.

4. **Follow the live activity**

   A **Computer** control appears near the chat input while access is active. Its status shows whether Bearly is starting, looking, or working in an app. The response also includes a **Computer Use** activity record with the app or screen being used and the actions involved.

   Keep an eye on the computer while Bearly works. You can move sensitive information out of view or stop the task at any time.

5. **Stop the task or revoke access**

   Select **Stop** in the **Computer** control or use the chat's stop control to stop the active task. Your approvals keep their selected duration. To remove access, open **Permissions** near the chat input (or **Full access** while active), choose **Standard**, then revoke Computer Use or select **Revoke all saved approvals**.



## Confirm sensitive actions

Some actions produce a second **Confirm this computer action** request. Review the reason, then select **Allow** only if the action matches what you asked for. Select **Deny** to prevent that action; you can then give Bearly a safer or more specific instruction.

For consequential work, tell Bearly where to pause—for example, before sending a message, submitting a form, saving over a file, or confirming a purchase. Computer Use can make mistakes, just as a person can when working quickly through an interface.

## Allow system access on macOS

Computer Use needs permission to observe the screen and control other apps. If Bearly asks for either permission:

1. Open **System Settings** → **Privacy & Security** → **Accessibility**, then allow **Bearly**.
2. Under **Privacy & Security**, open **Screen Recording** and allow **Bearly**. Some macOS versions call this **Screen & System Audio Recording**.
3. Return to Bearly and send the request again.

If your system asks you to reopen Bearly after changing a permission, reopen it before retrying the task.

## Finish Windows setup

On Windows, Bearly may offer a small Computer Use helper for more reliable control of protected and background windows.

- When **Finish setting up Computer Use** appears, select **Finish setup** and approve the Windows administrator prompt.
- When **Finish updating Computer Use** appears, select **Update** and approve the prompt.
- To check or retry setup later, open your account menu in the sidebar footer, choose **App Settings**, then select **Runtime** in the **Desktop** section. Review **Windows Computer Use helper** and choose **Install** or **Update**.

Windows asks for administrator approval because the helper is installed for everyone on that PC. Bearly itself continues to run as your normal user. You can choose **Remove Computer Use helper** from the same settings area; removal also requires administrator approval.

Computer Use may still handle ordinary visible windows without the helper, but protected or background windows can be less reliable. If a task stalls on one of those windows, finish or update the helper and try again.

## If Computer Use does not start

| What you see | What to do |
| --- | --- |
| Bearly says Computer Use is available only in the desktop app | Open the same conversation in the Bearly desktop app. Earlier results remain in the chat. |
| Bearly says the desktop app is too old or does not support Computer Use | Update Bearly to the current desktop release, reopen it, and retry. |
| Bearly asks for **Accessibility** or **Screen Recording** | Grant the requested macOS permission, then retry the message. |
| Bearly cannot find the app or window | Open or restore the relevant window so it is visible, then ask Bearly to continue. |
| A Windows helper installation or update was canceled | Open **App Settings** → **Runtime** and select **Install** or **Update** again. |
| Computer Use is blocked for your workspace | Ask a organization administrator whether the assigned policy permits Computer Use and the models it needs. |
| An action fails or the wrong window is active | Select **Stop**, bring the intended window into view, and send a narrower instruction. |

Administrators of managed organizations control availability from **Organizations** → organization name → **Policies** → policy name → **Code & Automation** → **Computer Use**. On means members can use it.

---

---
title: "Browse beside a Bearly chat"
description: "Open sites in Bearly Browser, annotate a page, or approve limited site access for a chat task."
canonical: "https://bearly.ai/docs/features/browser/"
markdown: "https://bearly.ai/docs/features/browser.md"
---

# Browse beside a Bearly chat

HTML: [https://bearly.ai/docs/features/browser/](https://bearly.ai/docs/features/browser/)

> Open sites in Bearly Browser, annotate a page, or approve limited site access for a chat task.

> **Note — Desktop only**
> Bearly Browser is available in the Bearly desktop app. It does not run on web or mobile. A team admin can disable Browser for managed accounts.


Bearly Browser lets a chat work on sites you approve, usually in the background. Open the Browser panel to watch, browse yourself, or point Bearly at part of a page.

The Browser uses a profile on this device for the current Bearly account. Its site sign-ins are separate from your normal browser, another Bearly account, and Bearly's own sign-in. To work in your everyday Chrome and its existing sign-ins instead, see [Control Chrome from Bearly](https://bearly.ai/docs/features/chrome-control.md).

## Turn on Browser



1. Open Bearly desktop and select **App Settings**.
2. Under **Desktop**, select **Browser**.
3. Turn on **Enable Browser**.
4. If you want Bearly to operate pages for chat tasks, also turn on **Allow chats to use Browser**. Standard asks before accessing new sites; [Full access](https://bearly.ai/docs/settings/agent-permissions.md) can use enabled Browser tools without asking each time.



**Block ads and trackers** is available in supported desktop versions. If a page stops working correctly, turn it off temporarily and reload the page.

## Browse manually

Select the **Open Browser** globe in the chat title bar. It shows the number of agent-owned tabs when any are open.

Enter a web address under **Browser address** and press <kbd>Enter</kbd>. The Browser opens beside the conversation with familiar **Back**, **Forward**, **Reload**, tab, and close controls. Use **Expand Browser** when you need more room, or **Open in system browser** to hand the current page to your usual browser.

You can keep up to eight Browser tabs open across your chats. Closing the panel keeps those tabs running in the background. Switching chats or pages in Bearly keeps them open too.

HTTP and HTTPS sites are supported, including local development servers such as `http://localhost:3000` and devices on your local network. HTTP connections are not encrypted.

### Review a certificate warning

If a site has an invalid or self-signed HTTPS certificate, Browser opens a **Your connection isn't private** warning—even when an agent opened the site in the background. Read **Certificate details** and choose **Go back**, or **Proceed anyway** only if you trust the site and network.

Only you can make this decision. An agent cannot approve the warning or bypass it. An exception is temporary and applies only to the exact origin, certificate, and reported certificate error; it does not change trust in your system browser. If an approval expires, you can still choose **Go back**, or reload to request a new warning.

## Ask Bearly to work with a site

With **Allow chats to use Browser** on, make the target and boundary explicit:

```text copy
In Browser, open the pricing page for this service and compare the three plans in a table. Do not sign in, start a trial, or submit anything.
```

In Standard, at **Let this chat use this site?**, review the site address and reason, choose **Once**, **This response**, **This session**, or **This chat**, then select the Allow button. You can also deny access or stop the task. Session approvals end when the chat reloads or the app restarts. Some settings or account-policy changes also reload the chat. Chat approvals cover future responses in this chat on the same device until revoked. Full access skips eligible tool approvals; certificate warnings and other required safety checks still apply.

An approval applies to the named site for that chat, not the whole web. Another site asks separately. A manual address you open while working in a chat is approved for that chat's Browser session.

Each chat's agent controls only its own tabs, marked with a robot icon. Tabs you create with the new-tab button are yours. Tabs share this account's Browser sign-ins; tab ownership does not create a separate signed-in account for each chat.

Use **Take over this tab** to pause its chat's agent while you work on the page. Select **Resume this tab's agent** when you are ready to hand control back. Taking over one chat's tabs does not pause other chats.

Submissions, opaque visual actions, website tools, and file downloads have separate site-specific approvals. Choose a longer duration to allow a batch of those actions. Bearly still reveals the relevant tab before submissions or opaque visual actions. Passwords, payments, purchases, account creation, CAPTCHAs, file uploads, permission changes, and destructive actions remain for you to complete directly on the visible page.

> **Caution — Treat pages as untrusted**
> A website can contain misleading text or instructions aimed at an AI assistant. Review the visible site, the requested access, and any proposed submission. Do not put a secret into the chat or approve an action merely because a page asks for it.


### Let a routine use Browser

A [routine](https://bearly.ai/docs/features/routines.md#let-a-routine-use-a-browser) can use Bearly Browser on its schedule when you turn on its **Browser access** and list the sites it may use. That setup is the approval: on those sites, the routine doesn't ask for access. Without **Click, type, and submit**, it only reads pages and follows links; with it on, it doesn't confirm submissions either and takes the actions its instructions call for. It can't open any other site, a page can't send it to one, and it never clicks or types into password or payment fields. It works in its own tabs with the Browser panel closed, uses this account's Browser sign-ins, and closes its tabs when the run ends. If it reaches an invalid certificate, it closes that tab instead of asking you.

## Point to part of a page

Use an annotation when a visual element is easier to identify than to describe:



1. Open the page and select **Capture and annotate page**.
2. Click an element, or drag over a region.
3. Under **What should Bearly change or inspect?**, add a short instruction.
4. Select **Attach**. The captured annotation is added to the chat composer.
5. Add any remaining context and send the message.



If the page changes before the annotation is attached, select the element again. Press <kbd>Esc</kbd> to cancel annotation mode.

## Bring in existing site sign-ins

Under **App Settings** → **Desktop** → **Browser**, **Import browser cookies** can copy current sign-ins from a supported Chrome or Edge profile.

1. Choose a **Browser profile** and select **Import**.
2. Review **Import cookies from _profile_?**, then select **Import cookies**.
3. Reload any open pages after the import finishes.

Only cookies are copied into the current account's Bearly Browser profile; the source browser is unchanged. Passwords, history, bookmarks, and autofill are not imported. Protected cookies may be skipped, and macOS may ask for access to the source browser's secure storage.

## Reset tabs or site data

The two reset actions have different consequences:

- **Reset open tabs** → **Reset tabs** closes Browser tabs and clears their local history and site approvals. Cookies and site sign-ins remain.
- **Delete browser data** → **Delete data** → **Delete browser data** signs this Bearly account out of sites and deletes its Browser cookies, cache, and local site storage on this device. It does not delete Bearly chats, Bearly account data, Browser preferences, or another Bearly account's Browser data.

Turning off **Enable Browser** stops the embedded pages and agent control. Re-enabling it reloads the retained tab addresses; unsaved page state may be lost. Site data is not deleted.

## If Browser does not work

| What you see | What to do |
| --- | --- |
| **Browser** is missing | Confirm that you are using the latest Bearly desktop app. If you use a managed account, ask your team admin whether Browser is disabled. |
| A Browser setting asks for an update | Install the latest Bearly desktop version, restart it, and return to **App Settings** → **Desktop** → **Browser**. |
| **That address uses an unsupported protocol.** | Enter a normal `https://` or `http://` web address. |
| A site breaks with blocking enabled | Turn off **Block ads and trackers**, then reload the page. |
| **Your connection isn't private** | Review the certificate warning yourself. The agent must wait for your decision. |
| The agent says you took over | Select **Resume this tab's agent** in Browser when you want the chat to continue. |
| A page takes too long to load | Agent navigation stops after 10 seconds. Review the current page before retrying; a form may already have been submitted. |
| Cookies cannot be imported | Close the source browser, make sure its profile is available on this device, and try again. Protected cookie storage can prevent some or all cookies from being copied. |
| Bearly cannot attach to a tab or annotation | Reload the page or open it in a new Browser tab, then retry the action. |
| You no longer want site access active | Open **Permissions** near the chat input, select **Standard** if Full access is active, then select **Revoke** or **Revoke all saved approvals**. This stops current work and asks again before further access. |

---

---
title: "Use Bearly on a web page"
description: "Summarize, question, or explain the page you are reading with the Bearly extension for Chrome, without leaving the tab."
canonical: "https://bearly.ai/docs/features/browser-extension/"
markdown: "https://bearly.ai/docs/features/browser-extension.md"
---

# Use Bearly on a web page

HTML: [https://bearly.ai/docs/features/browser-extension/](https://bearly.ai/docs/features/browser-extension/)

> Summarize, question, or explain the page you are reading with the Bearly extension for Chrome, without leaving the tab.

The Bearly extension puts a side panel next to the page you are reading. Summarize the page, pull out action items, explain a passage you selected, or ask a question and get an answer grounded in that page.

The extension also lets the Bearly desktop app work in your regular Chrome tabs. See [Control Chrome from Bearly](https://bearly.ai/docs/features/chrome-control.md) for that.

## Install and connect



1. Install **Bearly** from the [Chrome Web Store](https://chrome.google.com/webstore/detail/bearlyai/lkennebljmnnpimcgdjlocbncpamgipp).
2. Select the Bearly icon in the toolbar. The side panel opens.
3. Select **Sign in with Bearly** and finish the usual Google, Microsoft, or email login. Bearly shows which account will be connected before you continue.
4. Select **Allow website access**, then approve the browser's request for access to all websites.



The panel shows **Ready for this page** when it is set up. A small Bearly page button appears at the bottom-right edge of ordinary web pages. You stay connected in that browser profile until you disconnect.

## Work with the page

Open the page, then select the Bearly toolbar icon to open the panel. The presets sit above the composer:

- **Summarize** returns an executive summary and the key takeaways. After a summary, **More detail** asks for a longer version.
- **Action items** finds tasks, owners, deadlines, decisions, and open questions.
- **Explain selection** explains the text you selected first. It stays disabled until there is a selection.
- The **Ask this page** box at the bottom answers a question from that page rather than from general knowledge. Press <kbd>Enter</kbd> to send.

The Bearly page button runs the same three presets without opening the panel first. Select it, then choose **Summarize page**, **Action items**, or **Explain selection**; the answer arrives in the panel.

Answers appear in the panel with the page title and site they came from. Select **Copy** to take a result with you.

Follow-up questions continue in the same panel and can use the recent turns from that page. Select the **Clear conversation** eraser in the panel header to drop those turns. Moving to a different page starts fresh, and a **New page** divider marks the change.

### Google Workspace and YouTube

**Summarize**, **Action items**, and **Ask this page** work in Google Docs, Sheets, Slides, and Drive file pages using the document text your signed-in browser can already download. You do not connect Google Drive separately. On a YouTube video page, Bearly uses the video's transcript when one is published.

If your organization blocks document downloads, Bearly shows **Bearly cannot read this page**. **Explain selection** can be unavailable in Google Docs because the editor does not always expose the selection to the browser.

> **Note — What Bearly sends**
> Bearly sends nothing until you choose an action. It then sends a cleaned copy of the readable page text — or only your selection, for **Explain selection** — together with a picture of the visible part of the tab, which the model uses as supporting context. That picture can include anything on screen at the time, so close or scroll away from what you do not want included.
>
> Results and page text are not saved to your Bearly account. Recent turns stay in the browser only while that tab is open.


## Settings

Select the gear in the panel header.

- **Response language** sets the answer language. **Automatic** matches the page, or your question when you ask one.
- **Quick actions** chooses which presets appear above the composer and their order. **Ask this page** is always available, and at least one preset stays on.
- **Website access** turns page actions on or off across websites. **Page button** shows or hides the on-page Bearly shortcut, and a third row hides it for the site you are on.
- **Account** has **Open Bearly account** and **Disconnect Bearly**. Disconnecting removes Bearly from this browser and revokes its website access; your Bearly account and your work are unaffected.

You can also manage the page button from the page itself. Open it and select **Hide on this page** to hide it until you reload, **Don’t show on this site** to keep it hidden there, or **Hide Forever** to remove it from every website. **Hide Forever** asks for confirmation, and **Page button** in settings turns it back on.

## If something does not work

| What you see | What to do |
|---|---|
| **Connect Bearly before running an action.** | Select **Sign in with Bearly** and finish the browser login window. |
| **Bearly could not finish signing in. Try again.** | Return to the panel and select **Sign in with Bearly** again. |
| **Bearly cannot read this page.** | Open a normal `http` or `https` page. Browser settings pages, extension-store pages, and other protected surfaces do not allow extensions to read them. |
| **Bearly does not have permission for this site.** | Open the panel settings and turn on **Website access**, then approve the browser's request. |
| **Select text first.** | Select text in the main part of the page before running **Explain selection**. |
| **Bearly could not reach the AI service.** | Check your connection and run the action again. |
| The page button no longer appears | Open settings and check **Page button**, and the row for the current site. If **Website access** is off, turn it on first. |

If the extension keeps refusing requests, select **Disconnect Bearly** in settings, then sign in again. Write to [Bearly Support](mailto:support@bearly.ai) if the problem continues.

---

---
title: "Control Chrome from Bearly"
description: "Let a Bearly desktop chat read, navigate, and inspect your regular Chrome tabs, including sites you are already signed in to."
canonical: "https://bearly.ai/docs/features/chrome-control/"
markdown: "https://bearly.ai/docs/features/chrome-control.md"
---

# Control Chrome from Bearly

HTML: [https://bearly.ai/docs/features/chrome-control/](https://bearly.ai/docs/features/chrome-control/)

> Let a Bearly desktop chat read, navigate, and inspect your regular Chrome tabs, including sites you are already signed in to.

With the Bearly extension installed, a chat in the Bearly desktop app can work in the Chrome you already use: open and switch tabs, read pages, click and type, and inspect console and network activity. It uses your normal Chrome profile, so sites you are signed in to stay signed in.

> **Note — Desktop app and Chrome**
> This works in the Bearly desktop app for macOS, Windows, and Linux, with Chrome. It is not available on the web, on mobile, or in another browser. A organization admin can turn it off for managed accounts.


Bearly can work in a browser three ways:

- **Chrome control** uses your everyday Chrome and the sites you are already signed in to.
- [Bearly Browser](https://bearly.ai/docs/features/browser.md) is a separate browser beside the chat, with its own profile and its own sign-ins.
- [The extension side panel](https://bearly.ai/docs/features/browser-extension.md) answers questions about the page you are reading, without the desktop app.

## Connect Chrome



1. Install **Bearly** from the [Chrome Web Store](https://chrome.google.com/webstore/detail/bearlyai/lkennebljmnnpimcgdjlocbncpamgipp).
2. Open the Bearly desktop app and go to **App Settings** → **Desktop** → **Chrome Extension**.
3. Keep Chrome and the desktop app open. The status updates on its own and reads **Connected to Chrome** when Bearly is ready.



Nothing else to configure. Chrome and Bearly connect locally on your machine, and the connection survives restarting either one. You do not need to sign the extension in to Bearly for this — that sign-in is only for the extension's own side panel.

## Ask for a browser task

Say what you want and name Chrome. Bearly picks the tab and the tools.

```text copy
In Chrome, open the invoice list in my billing tab, find every invoice still unpaid after 30 days, and give me a table with vendor, amount, and due date. Don't pay or cancel anything.
```

Bearly can list your open tabs, so you can also point at one you already have: *"read the Chrome tab with the pull request and tell me what changed"*.

## Approve what Bearly can do

In Standard, Bearly asks in the chat before it touches Chrome, at one of three levels. [Full access](https://bearly.ai/docs/settings/agent-permissions.md) lets the chat use enabled Chrome tools without asking each time:

| Prompt | What it allows |
|---|---|
| **Read your Chrome tabs** | Tab list, page text, screenshots, console output, and page events |
| **Control Chrome** | Everything above, plus opening and closing tabs, navigating, clicking, typing, scrolling, and key presses |
| **Use Chrome DevTools** | Everything above, plus network requests and responses, storage, running JavaScript, debugging, profiling, and tracing |

Select **Allow** for the single action, or **Allow for…** to keep the level for **This response** or **For now**. **Deny** stops the action and lets the chat continue without it. Approving a higher level covers the lower ones for as long as the grant lasts.

While Bearly is attached to a tab, Chrome shows its own debugging banner. Closing the chat, stopping the response, or quitting either app detaches Bearly from every tab.

> **Caution — This is your real browser**
> Bearly acts in your signed-in Chrome profile and can see whatever those tabs can see, including account pages and, at the DevTools level, request contents and stored session data. Page text, console output, and network data are also untrusted input: a page can contain text written to mislead an assistant. State limits in your request, keep an eye on the tab, and use [Bearly Browser](https://bearly.ai/docs/features/browser.md) instead when a task does not need your own sign-ins.


## Turn it off

- Remove the grant for the current work by finishing or stopping the response. A **For now** grant ends when the chat session does.
- Quit the Bearly desktop app to disconnect Chrome entirely. Chrome control is unavailable until it is running again.
- Disable or remove the Bearly extension in Chrome to end it from the browser side. The extension side panel goes away with it.

## If Chrome control does not work

| What you see | What to do |
|---|---|
| **Chrome extension not connected** | Install or enable the Bearly extension in Chrome, then leave Chrome and the desktop app open until the status changes. |
| **Finishing secure connection…** stays on screen | Give it a moment. If it does not finish, restart Chrome, then the desktop app. |
| **Update Bearly** | Install the current desktop release and reopen it. |
| **Disabled by your workspace** | Ask a organization admin whether your assigned policy allows Chrome control. |
| Bearly never offers to use Chrome | Confirm you are in the desktop app, not the web app, and that the status reads **Connected to Chrome**. |
| **Chrome extension control could not complete that action.** | The tab may have closed, navigated, or lost the attachment. Ask Bearly to list tabs and try again. |
| Bearly is stuck on a sign-in wall or CAPTCHA | Finish that step yourself in the tab, then ask Bearly to continue. |

Administrators of managed organizations control availability from **Organizations** → organization name → **Policies** → policy name → **Code & Automation** → **Chrome extension browser**. On means members can use it.

---

---
title: "Find and check domain names"
description: "Ask Bearly for domain ideas around a product, company, or theme, then check whether a specific domain is still available to register."
canonical: "https://bearly.ai/docs/features/domain-research/"
markdown: "https://bearly.ai/docs/features/domain-research.md"
---

# Find and check domain names

HTML: [https://bearly.ai/docs/features/domain-research/](https://bearly.ai/docs/features/domain-research/)

> Ask Bearly for domain ideas around a product, company, or theme, then check whether a specific domain is still available to register.

Ask Bearly for domain ideas around a word, product idea, or naming brief, then have it check whether a specific domain is still available to register. Both happen in an ordinary chat.

Domain research works in Bearly chats on the web, desktop, iOS, and Android. It is not available to scheduled [Routines](https://bearly.ai/docs/features/routines.md). Bearly finds and checks domains; it does not buy or register them. You do that at a registrar.

## Ask for domain ideas



1. **Start the request**

   Type `/domains`, then select **Domain research** under **Research**. Asking in ordinary language works too, such as "suggest some domains for a plant care app."

2. **Say what the domain is for**

   Describe the product, audience, or theme. Add anything that narrows the result: endings you prefer such as `.com` or `.ai`, words that should influence the ideas, or a country whose local endings you want included.

   ```text copy
   /domains a scheduling tool for dental clinics. Prefer .com and .co, keep the name short, and work in ideas around "chair" and "slot".
   ```

3. **Read the ideas**

   Bearly answers with candidate domains. They are ideas, not an availability report — short, obvious domains are usually taken already. Pick the few worth checking.



## Check whether a domain is available

Ask Bearly about an exact domain:

```text copy
Is chairslot.com available to register?
```

Each check covers one domain and comes back one of three ways: the domain has no registration and is available, the domain is already registered, or the answer was inconclusive. Treat an inconclusive answer as unknown, never as available.

Shortlist before you check. A long list of ideas takes a separate lookup for every entry, and if you name several domains in one message, Bearly works through them one at a time.

Availability also changes minute to minute. Confirm the domain at the registrar you plan to buy from immediately before you pay.

## Name a product, company, or project

Naming a product, company, app, or project is one workflow this supports, and it goes better as a sequence than as a single question.



1. **Give Bearly the brief**

   Say what the thing does, who it is for, the tone you want, and any hard constraints: length, spelling you will not accept, endings you will consider.

2. **Ask for names, then domains around the survivors**

   Get a batch of candidate names first. Once a few feel right, ask for domain ideas built around those, including variants that read well as a domain — a short prefix or suffix, a different ending, or a two-word combination.

3. **Shortlist, then check the finalists**

   Run an availability check on the names you would actually use.

4. **Run the checks a domain cannot cover**

   See below before you commit to a name.



## What an available domain does not tell you

An available domain is not permission to use the name. Check these separately, and get professional advice before you build a business on a name:

- **Trademarks** in every country where you operate or sell.
- **Company and business registries** where you would register the entity.
- **App store names**, if you plan to ship an app.
- **Social handles** on the platforms you care about.
- **The open web** — an existing product, publication, or company already using the name.

Bearly can [search the web](https://bearly.ai/docs/features/web-search-deep-research.md) for existing use of a name, which is useful evidence. It is not a legal clearance, and neither is an available domain.

## If a check does not work

| What you see | What to do |
| --- | --- |
| **Domain research could not use those details. Check the domain or search terms.** | Give one complete domain with no scheme, path, or spaces: `example.com`, not `https://example.com/pricing`. |
| **Domain research is temporarily unavailable. Try again later.** | The lookup service did not answer. Wait, then ask again. |
| **Domain research could not complete that request. Check the details or try again later.** | Ask again. If it keeps failing, check the spelling of the domain. |
| Bearly reports the status as unknown or inconclusive | The registry answer was not conclusive. Ask again in a moment, and confirm at a registrar before acting on it. |
| `/domains` does not offer **Domain research** | Check for a Bearly update. On a managed account, ask a organization administrator whether the policy allows Domain research. |

Administrators of managed organizations control Domain research from **Organizations** → organization name → **Policies** → policy name → **Web & Research** → **Domain research**.

---

---
title: "Research keywords, rankings, and competitors with SEO Data"
description: "Pull keyword demand, live Google results, a site's search visibility, competitor gaps, backlinks, or a page audit into a chat, then keep working with the evidence."
canonical: "https://bearly.ai/docs/features/seo-data/"
markdown: "https://bearly.ai/docs/features/seo-data.md"
---

# Research keywords, rankings, and competitors with SEO Data

HTML: [https://bearly.ai/docs/features/seo-data/](https://bearly.ai/docs/features/seo-data/)

> Pull keyword demand, live Google results, a site's search visibility, competitor gaps, backlinks, or a page audit into a chat, then keep working with the evidence.

SEO Data brings current search data into a chat: how often people search for a keyword and how hard it is to rank for, who ranks on Google for a query right now, how visible a site is, where competitors outrank it, its backlinks, and how one page is built. The evidence stays in the conversation, so you can question it, combine it with your own files, and turn it into a brief or a plan.

SEO Data works in Bearly chats on the web, desktop, iOS, and Android, on every plan. It is not available to scheduled [Routines](https://bearly.ai/docs/features/routines.md) or to the experts in [Council of Experts](https://bearly.ai/docs/features/council-of-experts.md). Team administrators can turn it off with the **SEO Data** [team policy](https://bearly.ai/docs/teams/policies.md).

## Ask an SEO question

Ask in ordinary language. Bearly uses SEO Data on its own when a question needs current search data. To make sure it does, type `/seo` and select **SEO data** under **Research**.

```text copy
/seo Compare example.com with competitor-one.com and competitor-two.com in the United States. Find a content gap worth writing about, then show me who ranks for the best candidate on Google right now.
```

Results cover the United States in English unless you name another country, city, or language. Ask for mobile results when they matter; Google results default to desktop. SEO Data works one market at a time, so choose a country rather than "worldwide."

## What you can ask for

| Ask for | What Bearly can pull |
| --- | --- |
| **Keyword ideas** | Related keywords for up to five seed terms, with monthly search volume, ranking difficulty, cost per click, paid competition, search intent, and the yearly trend. |
| **Numbers for a keyword list** | The same metrics for a fixed list of up to 100 keywords. |
| **Live Google results** | Who ranks for up to five queries, the kind of result in each position, the search features on the page, and optionally where a given site ranks. |
| **A site's search visibility** | A site's estimated search visibility, the keywords it ranks for, its top pages, and its search competitors. |
| **A competitor comparison** | Your site against up to three competitors: overall visibility, keywords they rank for that you don't, or sites that link to them and not to you. |
| **Backlinks** | A site's backlink totals, referring domains, anchor text, and how many links it gained and lost each month over about the last year. |
| **A page audit** | One public page's status, title, description, canonical address, word count, links, size, and on-page problems such as a missing title or broken links. Ask for a performance check to add performance and SEO scores. |

Bearly writes the answer in the reply. To see exactly what came back, open the step in the reply's activity and look at **Result**.

## Read the numbers for what they are

- Search volume, difficulty, estimated traffic, and backlink counts are third-party estimates. They are not your analytics or Search Console data, and other tools may report different numbers.
- A missing value means the data isn't available, not that it's zero.
- Rankings describe the market, device, and depth that were checked. A page missing from the results may still rank elsewhere, on another device, or further down.
- A page audit reflects how the page looked when it was fetched. Performance scores are lab measurements under a simulated device and network, not data from real visitors.

Search operators such as `site:` or `intitle:` aren't supported. Describe what you want instead.

## What it uses

Each fresh request uses your plan allowance, based on the data pulled. There is no separate SEO fee. Keyword ideas and Google results make one request for each seed term or query, and a full site overview makes several, while numbers for a keyword list come back in a single request. If you repeat an identical request soon after in the same session, Bearly reuses the earlier data at no cost.

## If SEO Data can't answer

| What you see | What to do |
| --- | --- |
| **SEO data could not be retrieved. Try a narrower request or check the domain, URL, and location.** | Check that the domain or page address is public and spelled correctly, name a country, or ask for less at once. If you've used your plan allowance, this can also appear until it renews. |
| Part of a result is missing | Ask again for just that part, or narrow the request. |
| `/seo` doesn't appear | On a managed account, ask an administrator whether the **SEO Data** policy is on. |

---

---
title: "Find places and plan routes with Maps"
description: "Ask Bearly about places, reviews, directions, and travel times, and get cards with ratings, hours, routes, and stops along the way."
canonical: "https://bearly.ai/docs/features/maps/"
markdown: "https://bearly.ai/docs/features/maps.md"
---

# Find places and plan routes with Maps

HTML: [https://bearly.ai/docs/features/maps/](https://bearly.ai/docs/features/maps/)

> Ask Bearly about places, reviews, directions, and travel times, and get cards with ratings, hours, routes, and stops along the way.

Ask Bearly about places and getting around, and it answers with current map data: places that match what you want, a place's hours and ratings, what reviewers say, routes and travel times, and useful stops along a trip. Results appear as cards in the reply, with an interactive map where one helps.

Maps works in Bearly chats on the web, desktop, iOS, and Android, on every plan. It is not available to scheduled [Routines](https://bearly.ai/docs/features/routines.md). Team administrators can turn it off with the **Maps** [team policy](https://bearly.ai/docs/teams/policies.md).

## Find places

Ask in ordinary language. Name the area, or ask for places near you.

```text copy
Find three quiet coffee shops near Union Square that are open now and have outdoor seating.
```

Bearly uses Maps on its own when a question needs it. To make sure it does, type `/maps` and select **Maps** under **Research**.

The places card shows the first three matches, with **Show _N_ more** for the rest. Each row has the star rating and number of ratings, the price level, whether the place is **Open now** or **Closed**, the kind of place, and the address. Select a place to open it on a map in your browser.

You can narrow a search to places that are open now, have a minimum rating, or fall in a price range. For a search around one point, say the radius you want, up to 50 km (about 31 miles); the results then show each place's distance.

## Check one place

Ask about a specific place to see its details card: the address, **Open in Maps**, **Website**, the phone number, and **Hours**. Ask for amenities to see which apply, such as dine-in, delivery, takeout, reservations, outdoor seating, vegetarian options, or whether it suits children or groups.

**What people say** shows the number of ratings, a short summary of reviews with its source noted, and a sample of individual reviews. Select **View** on a review to read it in full, or **All reviews** to see every review.

## Read the reviews

For a fuller picture than the summary, ask Bearly to read the reviews:

```text copy
Read the reviews for Luna Bistro. What do people complain about most, and has that changed recently?
```

Bearly reads a broad mix of relevant, recent, and critical reviews and answers in the reply. The first batch can take about a minute.

## Get directions

Ask for directions between two places, with up to three stops in order. Choose **Driving**, **Walking**, **Cycling**, or **Transit**; driving is the default and accounts for current traffic.

```text copy
How long is the drive from the Ferry Building to Stanford if I leave at 4:30 pm tomorrow? Show me the alternatives.
```

The route card shows the **Best route** (for transit, **Recommended**) with its time and distance, **Open in Maps**, and any **Alternatives** with how much slower each one is. Transit routes list when to board, when you arrive, transfers, walking time, and the fare when it is known. Ask for turn-by-turn directions to add the steps.

You can set a departure time, or for transit, the time you need to arrive.

## Compare trips and find stops

- **Compare specific options.** Ask Bearly to compare two to four trips, such as the same commute by car and by train. The card names the fastest and shows how much slower the others are.
- **Measure many trips at once.** Ask for travel times between several starting points and destinations, up to 10 of each, such as from each teammate's neighborhood to three possible offices. Bearly answers in the reply.
- **Find stops along a trip.** Ask for a coffee stop, charging, or lunch on the way. The **Stops on the way** card lists places that need only a small detour, with **Directions** for the whole trip.

## How Bearly uses your location

Bearly looks up your location only when you ask about something near you without naming a place. A place you name always takes priority, and Bearly reuses a location already in the chat.

It asks your device first, so your device or browser may ask for permission. If your device's location isn't available, Bearly estimates an approximate location from your network connection, which sends your public IP address to a location service. Bearly then tells you which area it's using. If that's wrong, reply with the city or neighborhood you mean.

The location Bearly used becomes part of the chat's results, like any other result.

## Limits

- Distances and times are in miles, and results are in English. Distances under 1 km in a nearby search show in meters.
- Routes can't be set to avoid tolls, highways, or ferries.
- Maps reports current places and routes. It doesn't estimate foot traffic, demand, rent, or zoning.
- How complete the results are depends on the map data for the area, especially for transit.
- Each lookup uses your plan allowance. Reading reviews takes the longest.

## If Maps can't answer

| What you see | What to do |
| --- | --- |
| **No matching places found.** | Broaden the request: remove a filter, widen the area, or describe the kind of place differently. |
| **No route found.** | Check the start and end points, remove a stop, or try another travel mode. |
| A step says the map service is temporarily unavailable | Try again in a few minutes. |
| Bearly uses the wrong area for "near me" | Name the city or neighborhood in your next message. |
| `/maps` doesn't appear | On a managed account, ask an administrator whether the **Maps** policy is on. |

---

---
title: "Check weather, alerts, air quality, and pollen"
description: "Ask Bearly for current conditions, forecasts, public weather alerts, air quality, pollen, or past weather for a place, with results shown as cards in the chat."
canonical: "https://bearly.ai/docs/features/weather/"
markdown: "https://bearly.ai/docs/features/weather.md"
---

# Check weather, alerts, air quality, and pollen

HTML: [https://bearly.ai/docs/features/weather/](https://bearly.ai/docs/features/weather/)

> Ask Bearly for current conditions, forecasts, public weather alerts, air quality, pollen, or past weather for a place, with results shown as cards in the chat.

Ask Bearly about the weather for a place and it answers with current data: conditions right now, an hourly or daily forecast, active public weather alerts, air quality, pollen, or the weather on past dates. Most answers include a card with the numbers.

Weather works in Bearly chats on the web, desktop, iOS, and Android, on every plan. It is not available to scheduled [Routines](https://bearly.ai/docs/features/routines.md). Team administrators can turn it off with the **Weather** [team policy](https://bearly.ai/docs/teams/policies.md).

## Ask about the weather

Ask in ordinary language and name the place, or ask about where you are.

```text copy
Will it rain in Chicago on Saturday afternoon? We're planning an outdoor lunch.
```

Bearly uses Weather on its own when a question needs it. To make sure it does, type `/weather` and select **Weather** under **Research**.

Temperatures and speeds use Fahrenheit and miles per hour unless you ask for metric units. Results are in English.

## What you can ask for

| Ask for | What you get |
| --- | --- |
| **Current conditions** | A **Current conditions** card with the temperature, what it feels like, the conditions, and the chance of rain, wind, humidity, and UV index. |
| **An hourly forecast** | Up to 10 days ahead. The card shows 24 hours at a time in the place's local time; ask for more and Bearly continues. |
| **A daily forecast** | Up to 10 days, each with the conditions, chance of rain, and the high and low. |
| **Weather alerts** | Active official warnings that cover the place, with their severity, area, description, and when each expires. The card shows up to five alerts. |
| **Air quality** | The current air quality index and its category, the main pollutant, and pollutant levels. You can also ask for a forecast up to 96 hours ahead or readings from the last 30 days. |
| **Pollen** | Grass, tree, and weed pollen levels for up to five days. |
| **Past weather** | Daily weather for a date range of up to a year, hourly weather for up to 31 days, or up to 20 separate dates. Bearly answers in the reply without a card. |

Bearly checks for alerts when you ask about warnings or safety, or when the forecast points to a clear hazard. It does not check alerts with every forecast, so ask directly when it matters.

## How Bearly uses your location

When you name a place, Bearly uses it. When you ask about where you are, Bearly asks your device for its location first, so your device or browser may ask for permission. If your device's location isn't available, Bearly estimates an approximate location from your network connection, which sends your public IP address to a location service, and tells you which area it used. If that's wrong, reply with the city you mean.

The place Bearly used, with its coordinates, becomes part of the chat's results, like any other result.

## Limits

- A card shows the weather when it was fetched and doesn't update. Ask again for a newer reading.
- Alerts, air quality, and pollen aren't available everywhere. When a place isn't covered, Bearly says so.
- Each lookup uses your plan allowance. Past weather for separate dates counts each date as its own lookup.

## If Weather can't answer

| What you see | What to do |
| --- | --- |
| **Weather is unavailable for that location. Try a nearby city.** | Ask for the nearest larger city. |
| **Weather could not use those request details. Check the location or time window.** | Name the place more precisely, or shorten the time range. |
| **Weather is temporarily unavailable. Try again later.** | Wait a few minutes and ask again. |
| **Weather could not complete that request. Check the location and time window.** | Check the place and dates. If you've used your plan allowance, this can also appear until it renews. |
| `/weather` doesn't appear | On a managed account, ask an administrator whether the **Weather** policy is on. |

---

---
title: "Create interactive AI artifacts with Pages"
description: "Build, refine, reopen, and download interactive AI artifacts such as calculators, dashboards, planners, and small tools in Bearly."
canonical: "https://bearly.ai/docs/features/artifacts/"
markdown: "https://bearly.ai/docs/features/artifacts.md"
---

# Create interactive AI artifacts with Pages

HTML: [https://bearly.ai/docs/features/artifacts/](https://bearly.ai/docs/features/artifacts/)

> Build, refine, reopen, and download interactive AI artifacts such as calculators, dashboards, planners, and small tools in Bearly.

An interactive AI artifact is a file that an AI creates for you to use, not just read. Bearly calls these artifacts **Pages**. A Page is self-contained inside a conversation: describe what it should do in ordinary language, then use the result without having to write the code yourself.

## Create a Page


1. **Describe the outcome and the interaction**

   Tell Bearly what someone should be able to enter, choose, or explore. Include the important calculations, labels, and visual style.

   ```text copy
   Build a mortgage calculator with fields for the loan amount, interest rate, and term. Show the monthly payment, total interest, and a simple amortization chart.
   ```

   If the Page should use your data, select the plus icon → **Attach files**, or drag the files into the chat. Say which columns or values matter.

2. **Use the Page when it opens**

   Once the Page is ready, its preview appears in the conversation and usually opens in the side panel. Try its fields and controls there. Select **Open** on the preview whenever you want the larger view again.

3. **Refine it in the conversation**

   Ask for a precise change and refer to the Page by name when the chat contains several files.

   ```text copy
   In the mortgage calculator, add an optional annual overpayment and show how many months it removes from the loan.
   ```

   A successful revision creates a new version. In the open Page, use **Previous version** and **Next version** to compare revisions.


Pages work especially well for calculators, dashboards, interactive charts, planners, quizzes, small games, and clickable prototypes. For a data-heavy Page, explain the decisions it should help someone make—not only the chart types you want.

## Reopen or download a Page

You can return to a Page from either place:

- In the original conversation, select its preview and then **Open**.
- In the chat header, select the **Browse files** Files icon, choose **Generated**, then select the Page.

In the open Page, select **Download** to save the current version as an `.html` file. The downloaded file can be opened in a web browser.

To give someone a link they can open in their browser, [prepare and publish a share](https://bearly.ai/docs/chats/share.md). You can choose who has access and update the published version later.

> **Note**
> Choices you make inside a Page—such as entered form values or the current game state—are temporary. To keep a change to the Page itself, ask Bearly to revise it and wait for the new version.


## Understand the limits

A Page is one self-contained interactive file. It is a good fit for a focused tool, but not for a multi-page service with user accounts, a private database, or a long-running background process.

For calculations, data cleaning, or a one-off chart where the result matters more than the interface, use [Code Interpreter](https://bearly.ai/docs/features/code-interpreter.md). For a long draft that you want to edit as text, use [Canvas](https://bearly.ai/docs/features/canvas.md).

In a read-only conversation, you can open and download an existing Page, but Bearly cannot create a new version there.

## If the result is not what you expected

- If Bearly answers with ordinary text, reply with the controls and output you need and say, `Make this an interactive Page.`
- If a Page cannot load, retry the request or ask Bearly to create a fresh version. The earlier version remains available from the version controls.
- If a dashboard uses the wrong fields, attach the source file again and name the exact columns, units, and filters to use.

---

---
title: "Edit a long draft in Canvas"
description: "Turn a Bearly response into a Markdown draft, edit it directly, preview the formatting, and ask AI for focused revisions."
canonical: "https://bearly.ai/docs/features/canvas/"
markdown: "https://bearly.ai/docs/features/canvas.md"
---

# Edit a long draft in Canvas

HTML: [https://bearly.ai/docs/features/canvas/](https://bearly.ai/docs/features/canvas/)

> Turn a Bearly response into a Markdown draft, edit it directly, preview the formatting, and ask AI for focused revisions.

Canvas gives a long response its own editing space. You work with the Markdown text in the editor and see the formatted result in **Preview**.

## Create and open a Canvas


1. **Ask Bearly for a draft**

   Start with the content you want to edit—a memo, article, outline, proposal, or other long-form text.

   ```text copy
   Draft a concise project proposal with sections for the problem, approach, timeline, and risks.
   ```

2. **Turn the response into a Canvas**

   In the response controls, select **Edit in Canvas** (the notebook-and-pencil icon). Bearly turns the response text into a named Canvas and opens it.

3. **Return to it later**

   Select the Canvas card in the conversation, or select its **Open Canvas** control. Use **Close Canvas** or <kbd>Esc</kbd> to return to the full conversation.


## Edit and preview the draft

Type directly in the editor. Canvas uses Markdown characters for headings, emphasis, lists, links, and other formatting. For example:

```markdown copy
## Timeline

- **Week 1:** Research
- **Week 2:** Draft
- **Week 3:** Review
```

On a wide screen, the editor and formatted preview appear side by side. In a narrower window, use **Preview** to see the rendered document and **Edit** to return to the Markdown.

The toolbar also provides **Undo**, **Redo**, and **Copy**. **Copy** places the current Markdown on your clipboard so you can move the draft into another editor.

## Ask Bearly to revise text

For a focused edit, select the passage you want to change. In **Ask AI about this text…**, describe the revision, then select **Submit**.

```text copy
Make this paragraph more direct and preserve every date and number.
```

You can also keep the Canvas open and ask for a change in the conversation. Name the Canvas or quote the relevant sentence when more than one document could match. On a narrow screen, use **How can I edit this for you?** for a request about the open Canvas.

AI edits appear in the Canvas itself. Review them in **Preview** before continuing.

## Saving and recovery

Canvas saves direct edits after a short pause and again when you select **Close Canvas**. Wait a few seconds before closing the app if you have just typed a substantial change.

> **Caution**
> Canvas does not show a version browser. Before a major rewrite, use **Copy** to keep a separate copy of the Markdown if you may want to restore it later.


If a direct edit goes wrong, use **Undo** while the Canvas is still open. If an AI edit misses the mark, describe the correction or paste back the text you kept. If a document-storage error appears, copy any accessible text, restore your connection, and reopen the Canvas before continuing.

Canvas is for editable text. For an interactive calculator or dashboard, create a [Page](https://bearly.ai/docs/features/artifacts.md). For programmatic data analysis and generated files, use [Code Interpreter](https://bearly.ai/docs/features/code-interpreter.md).

---

---
title: "Analyze data with Code Interpreter"
description: "Let Bearly run Python for calculations, data analysis, charts, file conversion, and generated files."
canonical: "https://bearly.ai/docs/features/code-interpreter/"
markdown: "https://bearly.ai/docs/features/code-interpreter.md"
---

# Analyze data with Code Interpreter

HTML: [https://bearly.ai/docs/features/code-interpreter/](https://bearly.ai/docs/features/code-interpreter/)

> Let Bearly run Python for calculations, data analysis, charts, file conversion, and generated files.

Code Interpreter lets Bearly use Python when a question needs reliable calculation or file processing. It is useful for analyzing tables, creating charts, transforming files, and solving multi-step quantitative problems.

> **Note**
> A team administrator can disable Interpreter. If **Run Python** does not appear in the command menu, check your team's policy or ask an administrator.


## Analyze a file


1. **Attach the source material**

   Select the plus icon → **Attach files**, or drag the files into the chat. Code Interpreter cannot see files on your computer unless you attach them.

2. **Ask for the decision or output you need**

   Name the fields to use, the calculation to perform, and the form of the result. For example:

   ```text copy
   Analyze this sales spreadsheet. Compare revenue and gross margin by region, flag unusual month-over-month changes, and create a CSV with the underlying summary table.
   ```

   Bearly can choose Code Interpreter automatically when the task calls for it.

3. **Choose Run Python when you want to require it**

   Type `/` in the composer, then select **Run Python** under **Code**. Enter the task in **What would you like to run?** and send it.

4. **Review the run and its results**

   A **Run Python** step shows while the work is in progress. Select it to expand the **Input** and **Result**. If execution fails, the expanded step shows **Error** instead.

   Bearly explains the result in the conversation. Charts appear as previews, and generated spreadsheets, documents, images, or other files appear with the response.


## Open, download, and continue with results

Select a generated file to preview it, then use **Download** in the open file panel. You can also find it later through the **Browse files** Files icon in the chat header.

To continue working, stay in the same conversation and refer to the generated file by name:

```text copy
Using the regional-summary.csv you just created, add each region's share of annual revenue and sort from highest to lowest.
```

Each execution starts with a fresh workspace. Bearly can reuse an earlier output only by supplying that generated file to the next run, so naming the file removes ambiguity.

## Good tasks for Code Interpreter

- Calculate statistics, financial scenarios, or other multi-step math.
- Inspect, filter, join, summarize, or reshape CSV and spreadsheet data.
- Create charts and save the underlying data as a file.
- Convert or modify documents, images, audio, and other supported files.
- Generate a clean CSV, spreadsheet, document, image, or text file from computed results.

For an inline KPI, bar chart, or line chart whose source data stays attached, see [Create charts from spreadsheets](https://bearly.ai/docs/features/charts.md). For a persistent interactive interface, such as a calculator or dashboard that someone will operate, create a [Page](https://bearly.ai/docs/features/artifacts.md) instead.

## Limits and recovery

Code Interpreter runs away from your device. It cannot use your local applications, folders, or installed software, and it does not keep a working directory between executions. It can retrieve public internet resources when needed.

Runs have bounded time, memory, and computing power. Large datasets, intensive media processing, and long-running programs may fail. A single run can return up to five output files.

If a run fails:

- Expand **Run Python** and read **Error** for the immediate cause.
- Confirm every required input file is attached and readable.
- Ask Bearly to correct the code and rerun it; you do not need to write the correction yourself.
- Split a large dataset or multi-stage job into smaller tasks, saving an intermediate file between stages.

For consequential work, ask Bearly to show its assumptions and validation checks. Review the source data and generated file before relying on the result.

When the result needs to be a polished PDF, Word document, presentation, or spreadsheet rather than an analysis file, use [Document Editor](https://bearly.ai/docs/features/document-editor.md).

---

---
title: "Create charts from spreadsheets"
description: "Turn a spreadsheet or CSV into a KPI card, bar chart, or line chart with the source data ready to inspect and download."
canonical: "https://bearly.ai/docs/features/charts/"
markdown: "https://bearly.ai/docs/features/charts.md"
---

# Create charts from spreadsheets

HTML: [https://bearly.ai/docs/features/charts/](https://bearly.ai/docs/features/charts/)

> Turn a spreadsheet or CSV into a KPI card, bar chart, or line chart with the source data ready to inspect and download.

Attach a spreadsheet or CSV and ask Bearly to turn the relevant rows into a KPI card, bar chart, or line chart. The chart stays in the conversation with the summarized data it was drawn from.

Charts works in Bearly chats on the web, desktop, iOS, and Android. Bearly usually needs [Code Interpreter](https://bearly.ai/docs/features/code-interpreter.md) to summarize a spreadsheet before charting it. In the desktop app, you can instead use [Work locally](https://bearly.ai/docs/features/work-locally.md) and approve the required file and command actions.

## Create a chart



1. **Attach the source file**

   Select the plus icon → **Attach files**, or drag a spreadsheet or CSV into the chat. You can also use a file that is already in the conversation.

2. **Choose Chart**

   Type `/chart`, then select **Chart** under **Tools**. You can also ask for a chart in ordinary language when the source file and desired result are clear.

3. **Name the comparison or decision**

   Tell Bearly which values to calculate, how to group them, and which chart form you want. For example:

   ```text copy
   Compare quarterly revenue by region as a grouped bar chart. Use USD, sort the quarters in date order, and keep the summarized data with the chart.
   ```

   Bearly calculates the summary in code, saves it as a CSV, and draws the chart from that file. If you use **Work locally**, review and approve each requested file or command action.

4. **Check the chart and its data**

   The result appears in a card labeled **Chart**. Move the pointer across a bar or line chart to inspect the nearest value. On a touch screen, tap the plot to keep that value visible.

   Select **View data** to open the summarized rows behind the chart, then **View chart** to return. Every chart includes **Download CSV**. Bar and line charts also include **Download PNG**.



## Choose the right result

Charts supports these result types:

- **KPI card** — one value with an optional goal and progress indicator.
- **Bar chart** — one series, grouped series, or stacked series across categories.
- **Line chart** — one or more series across dates or ordered labels.

Ask for a table when you need row-level records. Ask for a pivot table when the grouped table itself is the result. Those stay as downloadable files instead of being forced into a chart.

For a dashboard, calculator, map, pie chart, scatter plot, or another interactive layout, ask Bearly for a [Page](https://bearly.ai/docs/features/artifacts.md) or use [Code Interpreter](https://bearly.ai/docs/features/code-interpreter.md) to create a separate file.

## Data limits

Bearly charts summarized data rather than a large raw export:

- A KPI uses exactly one summary row.
- A bar chart supports up to 100 rows and 30 categories.
- A line chart supports up to 2,000 rows and 12 series.
- The summarized CSV can contain up to 10 columns and must be 2 MB or smaller.

Bearly does not silently sample or cut off rows to make a chart fit. If the summary is too large or contains duplicate category-and-series pairs, it must rebuild the summary before trying again.

## If a chart does not appear

| What you see | What to do |
| --- | --- |
| `/chart` does not offer **Chart** | Check for a Bearly update. On a managed account, ask a organization administrator whether the policy allows Charts. |
| Bearly cannot prepare the spreadsheet | Confirm that the source file is attached and [Code Interpreter](https://bearly.ai/docs/features/code-interpreter.md) is available. In the desktop app, you can turn on [Work locally](https://bearly.ai/docs/features/work-locally.md) and approve the requested actions instead. |
| Bearly reports missing fields, duplicate values, or too much data | Ask it to rebuild a smaller summary with one row per category and series pair, then create the chart again. |
| **The chart could not load. The data is available below.** | Review or download the CSV shown in the fallback. Ask Bearly to create the chart again if the data is correct. |
| The chart says its source no longer exists | Attach the source file again and ask Bearly to rebuild the chart. |

Administrators of managed organizations control Charts from **Organizations** → organization name → **Policies** → policy name → **Media & Creativity** → **Charts**.

---

---
title: "Create and edit documents"
description: "Use Document Editor to create, revise, and convert PDF, Word, PowerPoint, and Excel files in a Bearly chat."
canonical: "https://bearly.ai/docs/features/document-editor/"
markdown: "https://bearly.ai/docs/features/document-editor.md"
---

# Create and edit documents

HTML: [https://bearly.ai/docs/features/document-editor/](https://bearly.ai/docs/features/document-editor/)

> Use Document Editor to create, revise, and convert PDF, Word, PowerPoint, and Excel files in a Bearly chat.

Document Editor turns a plain-language request into a new document or a revised copy of one you attach. It works with PDF, Word, PowerPoint, and Excel files, including common `.pdf`, `.docx`, `.pptx`, and `.xlsx` formats.

> **Note**
> A team administrator can disable Interpreter, which also removes **Document Editor**. If the command does not appear, check your team's policy or ask an administrator.


## Create, edit, or convert a file



1. **Choose Document Editor**

   Type `/` in the message box, then select **Document Editor** under **Documents**.

2. **Attach any source files**

   To revise, combine, extract from, or convert existing material, select the plus icon → **Attach files** and add every file Bearly should use. Skip this step when creating a document from scratch.

3. **Describe the finished file**

   In **What would you like to do with your document?**, name the output format, the intended reader, the outcome, and anything that must remain unchanged.

   ```text copy
   Turn the attached research notes into a six-slide PowerPoint for an executive review. Keep every quoted figure exactly as written, add a source note to each slide, and use the attached deck as the visual template.
   ```

   For an edit, distinguish required changes from material to preserve:

   ```text copy
   Revise the attached Word document for clarity and return a new DOCX with tracked changes. Preserve the headings, footnotes, tables, and existing comments.
   ```

4. **Review the work**

   Bearly may show a **Run Python** step while Document Editor is working. Select the step to inspect its **Result** or, if the task fails, its **Error**.

   Each completed document appears as a file in the conversation. Select it to preview the result when that format supports a preview.

5. **Download the file you need**

   In the open file panel, select **Download**. For some Word, PowerPoint, and Excel files, the preview is a converted PDF and the download menu offers **PDF** and **Original**. Choose **Original** when you need the editable source format.



The file you attached remains unchanged in the chat. Document Editor returns its work as a new file, so you can compare the two or ask for another revision.

## Be precise about what should survive

Document work is more reliable when the request names both the desired change and the details that matter:

- For a PDF, give page numbers and say whether to merge, split, rotate, extract, fill, watermark, or run text recognition on scanned pages. Check names, numbers, and tables carefully after text recognition.
- For a Word document, say whether to preserve styles, comments, footnotes, tables, or tracked changes. If you want tracked changes in the result, request them explicitly.
- For a PowerPoint, attach the template to follow and name the audience, slide count, aspect ratio, and visual elements to retain. Ask Bearly to check every slide for clipped or overlapping content.
- For an Excel workbook, identify the relevant sheets and columns. Say whether formulas, formatting, charts, comments, or worksheet names must remain, and ask Bearly to keep calculated values as formulas where appropriate.

A single run can return up to five files. If you need more, divide the work into a few clearly named batches.

## Start with a focused document task

When the input and result are already clear, these focused tools collect the material before opening the work in Bearly:

- [Convert Markdown to PDF](/tools/markdown-to-pdf) when the source is Markdown and the finished file should preserve headings, tables, links, and code blocks.
- [Summarize a PDF](/tools/pdf-summarizer) when you need a concise account of one document with page references.
- [Convert PDF to Excel](/tools/pdf-to-excel) when the goal is a reviewable spreadsheet rather than prose.
- [Compare PDFs](/tools/compare-pdfs) when you need the differences between two versions before revising either file.

Each tool starts a focused workflow and then opens the result in the Bearly workspace for review and follow-up.

## Verify the result before sharing it

The preview is useful for a first review, but open the downloaded file in the application where it will be used before treating it as final.

- Compare the page or slide count with the source and check fonts, spacing, images, tables, and page breaks.
- In a spreadsheet, recalculate the workbook and inspect important formulas, totals, dates, and external links.
- In a scanned PDF, compare critical names and figures with the original page image.
- Test interactive or advanced elements such as forms, comments, animation, embedded media, and macros. Complex files may need manual correction even when the text and data are accurate.

> **Caution**
> Document conversion can change layout, fonts, pagination, formulas, or advanced file behavior. Review consequential documents in their original application before signing, publishing, filing, or sending them.


## Continue or recover

Generated documents remain available through the **Browse files** Files icon in the chat header. Choose **Generated** to narrow the list.

To revise an output, stay in the same conversation and refer to the file by name:

```text copy
Using executive-review.pptx, shorten slides 3 and 4, keep the source notes, and return a new PowerPoint.
```

If a task fails:

- Expand **Run Python** and read **Error**.
- Confirm that the source file opens normally and attach it again if needed.
- Retry one conversion or revision at a time, especially for a large or complex document.
- Name the exact source file and requested output format.
- If the preview is unavailable or looks incomplete, download **Original** and inspect it in Word, PowerPoint, or Excel.

For calculations and data analysis that do not need a polished document, use [Code Interpreter](https://bearly.ai/docs/features/code-interpreter.md).

---

---
title: "Check grammar and clarity"
description: "Review grammar, spelling, punctuation, style, vocabulary, and clarity suggestions without losing your original meaning."
canonical: "https://bearly.ai/docs/features/grammar-correction/"
markdown: "https://bearly.ai/docs/features/grammar-correction.md"
---

# Check grammar and clarity

HTML: [https://bearly.ai/docs/features/grammar-correction/](https://bearly.ai/docs/features/grammar-correction/)

> Review grammar, spelling, punctuation, style, vocabulary, and clarity suggestions without losing your original meaning.

**Grammar Correction** marks possible writing problems inside your original text. You can inspect each explanation, accept only the changes you agree with, and copy the revised version when you are finished.

## Check a passage



1. **Choose Grammar Correction**

   Type `/grammar` in a chat, then select **Grammar Correction** from the command menu.

2. **Add the text**

   Paste or type the passage into the message box. For a useful review, include complete sentences and enough surrounding context to make the intended meaning clear.

3. **Send the message**

   Bearly returns the passage with possible issues underlined. The summary above it shows how many suggestions are still waiting for review—or **No issues found** when there are none.



Grammar Correction can flag grammar, spelling, punctuation, style, vocabulary, and clarity. An underline is a suggestion, not a verdict: names, quotations, house style, dialect, and deliberately unusual phrasing may be correct as written.

## Review the suggestions

Select any underlined passage to see:

- The kind of issue Bearly noticed.
- A short explanation.
- The original wording and proposed change.
- **Accept**, which applies that one change.

Choose **Accept All** only when you have reviewed the set and want every remaining correction. Leave any suggestion you do not want untouched; it will not be applied.

After changing the passage, select **Check Again** to run a fresh review on the current version. Use **Copy corrected text** to copy the complete passage, including accepted changes, to the clipboard.

> **Caution**
> For legal language, quotations, names, technical terminology, and carefully styled prose, review each suggestion individually. A grammatically smoother sentence can still change emphasis or meaning.


## Get a better review

- Ask for one purpose at a time. Grammar Correction is for sentence-level review; use an ordinary chat request when you want a rewrite for a new audience or tone.
- Break a very long document into meaningful sections so each passage keeps enough context.
- If a suggestion appears in the wrong place, copy the current text into a new **Grammar Correction** request and run it again.
- If **Check Again** fails, the accepted text remains in the card. Wait for the connection to recover, then select it again or copy the text into a new request.

---

---
title: "Generate images"
description: "Create images in chat, try models in parallel, and keep editing from any result."
canonical: "https://bearly.ai/docs/features/image-generation/"
markdown: "https://bearly.ai/docs/features/image-generation.md"
---

# Generate images

HTML: [https://bearly.ai/docs/features/image-generation/](https://bearly.ai/docs/features/image-generation/)

> Create images in chat, try models in parallel, and keep editing from any result.

Ask Bearly to create an image, then keep creating and editing in the image editor inside the conversation.

## Create your first image

Open a chat, describe the image you want, and send the message. Name the subject and any details that matter, such as the setting, style, lighting, or text to include.

```text copy
Create a square editorial illustration of a night train crossing a snowy landscape, seen from above. Use deep blue shadows, warm window light, and no text.
```

Bearly creates the image and opens the editor. Uploaded images, generations, and edits remain available in its thumbnail strip.

## Create more images and compare models



1. **Select Create image**

   Use **Create image** in the editor's action bar to start from a description without using a selected image as a reference.

2. **Describe the image**

   Enter your description in **Describe the image you want to create…**. Select **Refine** for a suggested revision, then **Use this** to accept it or **Dismiss** to keep your wording.

3. **Select a model to run**

   Each model button starts one image immediately. Select another model, or the same model again, to start another run while earlier ones finish.



**Fast** is a shortcut for quick drafts; **Good** favors quality. These are curated choices, and results and timing vary. Choose **All models** to run a specific model. **Recent** brings back compatible models you have used successfully; models already featured above are omitted.

Each run uses the description you submitted at that moment. You can keep changing your draft while results arrive in the strip. Each run uses AI credits; see [Billing and usage](https://bearly.ai/docs/account/billing-usage.md).

Select a completed thumbnail to use it for editing. **Create image** and editing keep separate drafts while you work. To repeat a selected result's original instructions, model, and inputs, select **Re-run**. The latest image response shows the editor. Earlier image responses show **Go to latest image edit**; select it to jump forward to the editor.

## Save or inspect a result

Choose **Edit image** and select the result's thumbnail. On a computer, move over the image to reveal its actions; on touch devices, the actions remain visible:

- **Copy image** copies the image to the clipboard.
- **Download** saves the image file.
- **View full size** opens a larger preview with zoom and previous/next controls. Press <kbd>←</kbd> or <kbd>→</kbd> on a keyboard to move through the image strip.

> **Caution**
> Generated images can contain inaccurate text, anatomy, logos, or small details. Inspect anything consequential at full size before publishing or presenting it.


## Manage unfinished runs

- Select a **Running** thumbnail, then **Cancel run**, to stop a run controlled by this editor.
- Select a **Failed** or **Canceled** thumbnail for its message and **Try again**. You can also change the instructions or choose another model for a new attempt.
- **Pending** means the editor cannot confirm whether the run is still active. **Dismiss** removes that entry without deleting saved images or stopping work elsewhere. A completed image can still appear later.
- If the message says the request was blocked, revise the description before trying again.
- If an image cannot be loaded, reopen the conversation or generate it again if the file is no longer available.
- Unavailable controls may mean the chat is read-only or your organization has disabled image tools. Ask the owner or administrator for access.

To change an existing image, see [Edit an image](https://bearly.ai/docs/features/image-editing.md). To use several sources together, see [Combine images](https://bearly.ai/docs/features/image-combining.md).

---

---
title: "Edit an image"
description: "Change an uploaded or generated image, select an area to edit, or extend its edges in the conversation."
canonical: "https://bearly.ai/docs/features/image-editing/"
markdown: "https://bearly.ai/docs/features/image-editing.md"
---

# Edit an image

HTML: [https://bearly.ai/docs/features/image-editing/](https://bearly.ai/docs/features/image-editing/)

> Change an uploaded or generated image, select an area to edit, or extend its edges in the conversation.

Use a written instruction to change an image, select a particular area, or add space around its edges. Each result is a new image; the source stays available.

To start in a new chat, select the plus icon beside the message box, choose **Attach files**, and add an image. Ask Bearly for the change you want and send the message. The result opens in the image editor inside the conversation.

## Edit from the image editor



1. **Choose Edit image and a source**

   Select **Edit image** in the action bar, then choose a thumbnail. Use **Upload** beside the strip to add another image. The selected source appears above the strip.

2. **Describe the change**

   Enter an instruction in **What would you like to change?** Name what should change and any details that should stay the same.

   ```text copy
   Replace the daytime sky with a restrained pink-and-blue sunset. Keep the building, camera angle, and all signs unchanged.
   ```

3. **Select a model to run**

   Select **Fast**, **Good**, a featured or recent model, or a model in **All models**. Each click starts a separate result. You can try more models while earlier runs finish.

4. **Choose a result to continue**

   New images appear in the strip without replacing your working source or instruction. Select a result's thumbnail when you want to use it for the next edit.



**Refine** suggests a revision to your instruction; **Use this** accepts it and **Dismiss** keeps your wording. **Re-run** repeats a selected result's original instructions, model, and inputs. Choosing a model button uses your current instruction and selected source instead.

## Change one area

Select **Change area** and choose the source image. Drag a rectangle over the area to change, or choose **Select center**. Enter the change you want, then select a compatible model to run it. To replace the selection, drag again; the × control clears it.

Choosing another image or switching actions clears the area selection. **All models** explains when a model does not support area edits.

## Extend the edges

Select **Extend edges** and choose the source image. The editor shows space around all four edges. Describe what should fill that space, such as continuing the landscape or adding more room around the subject, then select a compatible model.

This action extends all four sides together. The final dimensions depend on the model.

> **Caution**
> Editing regenerates the picture. Even an area or edge edit may change details you wanted to preserve. Check faces, hands, text, product marks, proportions, and the edges around changed areas before using the result.


## Inspect or recover an edit

Return to **Edit image** and select a result to use **Copy image**, **Download**, or **View full size**. On a computer, move over the image to reveal these actions; on touch devices, they remain visible. In the full-size preview, use the previous and next controls or press <kbd>←</kbd> and <kbd>→</kbd> to move through the image strip.

- To change the source, select another thumbnail. Area edits need a new selection afterward.
- If model buttons are unavailable, check that you have a source, an instruction, and an area selection when needed. Open **All models** for compatibility explanations.
- If too much changed, return to the source and give a narrower instruction that names the details to preserve.
- If a source cannot be loaded, use **Upload** to add it again.
- For a failed or canceled run, select its thumbnail and **Try again**, or start a new attempt with another model. See [Manage unfinished runs](https://bearly.ai/docs/features/image-generation.md#manage-unfinished-runs) for cancellation and pending entries.
- If editing is unavailable throughout the chat, it may be read-only or image tools may be disabled by your organization.

To use several sources in one result, see [Combine images](https://bearly.ai/docs/features/image-combining.md). To start from a description, choose **Create image** or see [Generate images](https://bearly.ai/docs/features/image-generation.md).

---

---
title: "Combine images"
description: "Select multiple source images, describe how their elements should relate, and create a new composition in chat."
canonical: "https://bearly.ai/docs/features/image-combining/"
markdown: "https://bearly.ai/docs/features/image-combining.md"
---

# Combine images

HTML: [https://bearly.ai/docs/features/image-combining/](https://bearly.ai/docs/features/image-combining/)

> Select multiple source images, describe how their elements should relate, and create a new composition in chat.

Use several images as references for one new composition—for example, placing a product in a scene or bringing subjects from separate pictures together.

To start in a new chat, select the plus icon beside the message box, choose **Attach files**, and add at least two images. Describe how you want to combine them and send the message. Bearly creates a result and opens the image editor.

## Combine sources in the editor



1. **Select Combine images**

   Choose **Combine images** in the action bar. Any source already selected stays selected.

2. **Choose at least two sources**

   Select thumbnails to add or remove images. The selected images appear above the strip, and the caption shows the selection count. Use **Upload** to add another reference.

3. **Explain each source's role**

   Describe what each image contributes and where those elements should appear. Refer to visible content when several sources look similar.

   ```text copy
   Use the room as the setting. Place the green armchair from the other image beside the window, at a believable scale. Keep the room's camera angle and evening light.
   ```

4. **Select a model to run**

   Each model button starts a separate result. Choose a featured or recent model, or open **All models**. Models that cannot use the selected number of images are unavailable, with an explanation in the menu.



The result appears in the strip, and your source selection stays in place so you can try another model or instruction. The source files remain unchanged.

To edit the composite on its own, choose **Edit image**, then select the result's thumbnail. In **Combine images**, clicking a result adds or removes it from the source group.

## Refine the composition

Name which source provides the setting, which objects or subjects to bring in, and what should remain unchanged. If the sources disagree in perspective or lighting, say which one should take priority.

> **Caution**
> Combining images is a generative edit. Bearly may reinterpret a face, product, logo, texture, or background while building the scene. Review identity-sensitive and brand-sensitive details closely.


- If the wrong sources are selected, select their thumbnails again to remove them, then add the intended sources.
- If a model is unavailable, reduce the source count or choose another model in **All models**. Each model has its own input limit.
- For a crowded result, combine two images first, then use that result with the next source.
- If a source cannot be loaded, upload it again and replace it in the selection.
- For failed, canceled, or pending runs, see [Manage unfinished runs](https://bearly.ai/docs/features/image-generation.md#manage-unfinished-runs).

For area edits or extending a single image, see [Edit an image](https://bearly.ai/docs/features/image-editing.md).

---

---
title: "Create and edit videos"
description: "Generate a short video from text or an image, request sound and format choices, and transform an existing clip."
canonical: "https://bearly.ai/docs/features/video-generation/"
markdown: "https://bearly.ai/docs/features/video-generation.md"
---

# Create and edit videos

HTML: [https://bearly.ai/docs/features/video-generation/](https://bearly.ai/docs/features/video-generation/)

> Generate a short video from text or an image, request sound and format choices, and transform an existing clip.

Bearly can create a short video from a written scene, animate a still image, or transform an existing clip. You describe the motion and format in the chat; the finished MP4 appears in an inline player.

Video tools are available in Bearly chats unless a managed-team policy disables them. Generation uses your account's AI credits, and higher resolutions cost more.

## Generate a video



1. **Choose Generate Video**

   Type `/`, then select **Generate Video**.

2. **Optionally attach an image**

   For text-to-video, no image is required. To animate a still image, select the plus icon beside the message box, choose **Attach files**, and add it before sending the request.

   You can attach both an opening and closing image, but identify which should be the first frame and which should be the final frame. A final-frame image cannot be used without a starting image.

3. **Describe the scene and motion**

   Include the subject, action, camera behavior, light, and pacing. Add format choices only when they matter.

   ```text copy
   Create an 8-second vertical video of this paper kite lifting into a coastal wind. Start with a close shot, then pull back slowly as it rises. Use 1080p and include quiet wind and surf, with no speech.
   ```

4. **Send the request**

   Bearly generates the clip and places it in the conversation. The result shows the prompt and useful details such as duration, resolution, and whether audio was included.



If you omit format choices, Bearly uses a 5-second duration, 1080p resolution, and a 16:9 frame. You can request:

- A duration from 3 to 15 seconds.
- 720p, 1080p, or 4K resolution.
- A 16:9 landscape, 9:16 vertical, or 1:1 square frame.
- Generated sound effects, music, or speech.

> **Note**
> Generated audio is available when creating a new video. When editing an existing clip, you can keep or remove its original sound, but the edit command does not add a newly generated soundtrack.


## Edit an existing video



1. **Choose the source clip**

   Use a video already in the chat or attach a short MP4 with **Attach files**.

2. **Choose Edit Video**

   Type `/`, select **Edit Video**, and make sure the intended source is attached or clearly identified.

3. **Describe the transformation**

   Explain what should change and what should remain. The source sound is kept unless you ask Bearly to remove it.

   ```text copy
   Keep the camera movement and timing, but change the scene from afternoon to blue hour. Preserve the original sound.
   ```

4. **Review the new clip**

   Bearly adds the edit as a separate video marked **Edited**. The original clip remains unchanged.



Use a short source clip for editing. If Bearly rejects its length, trim it to 3–10 seconds. Uploaded source videos for editing must be no larger than 50 MB.

## Play and save the result

The inline player includes **Play video**, **Mute video**, playback speed, and **Download video** controls. Downloads use the MP4 format. The video also remains in the chat for later follow-up work.

To save an existing video from a supported public site instead of generating one, see [Download a video](https://bearly.ai/docs/features/download-video.md).

> **Caution**
> Generated video can introduce inconsistent faces, hands, objects, text, motion, or audio. Review the entire clip—including its final frames and sound—before publishing it or using it as evidence.


## If video generation or editing fails

- If an attached image or video cannot be loaded, attach it again and resend the request.
- If an edit says the source is too large, use a shorter or lower-resolution clip under 50 MB.
- If Bearly cannot read the source duration, re-export or re-upload the clip, then try again.
- For a failed generation, retry with a simpler motion description. Choose a lower resolution when you want to use fewer credits.
- If the video command is missing, ask a team administrator whether your assigned policy allows video tools.
- If Bearly reports a usage limit, open [Billing and usage](https://bearly.ai/docs/account/billing-usage.md) to review your available credits.

---

---
title: "Download a video"
description: "Save videos from supported public sites to a Bearly chat, follow download progress, and keep the resulting files."
canonical: "https://bearly.ai/docs/features/download-video/"
markdown: "https://bearly.ai/docs/features/download-video.md"
---

# Download a video

HTML: [https://bearly.ai/docs/features/download-video/](https://bearly.ai/docs/features/download-video/)

> Save videos from supported public sites to a Bearly chat, follow download progress, and keep the resulting files.

Download Video saves a video from a supported link and places the playable file in your Bearly conversation. It works with sites such as X, Vimeo, TikTok, Reddit, and Twitch; site support can change and is not universal.

> **Note — YouTube requires Bearly desktop**
> **Download Video** supports YouTube in the Bearly desktop app. Outside desktop, the command supports compatible non-YouTube sites, but not YouTube.


**Download Video** isn't available in a read-only chat, or when a managed organization's policy turns off **File uploads** (under **Privacy & Security**), because it saves a file to the chat.

## Download from a link



1. **Choose Download Video**

   Type `/` in the message box, then select **Download Video** under **Media Tools**.

2. **Add the video address**

   In **Provide the URL of the video you want to download**, paste the address of the individual video page. You can include up to three URLs in one request; put each on its own line so Bearly can distinguish them.

3. **Send the request and follow progress**

   The **Downloading video** step shows which item Bearly is handling and how many have finished. On desktop, it may also show preparation, download percentage, and file-transfer progress. Other platforms may show fewer intermediate details.

   While Bearly is working, the send button becomes **Stop**. Select it to cancel the active request.

4. **Play or save each result**

   Each successful download appears in the chat as soon as it is ready. Play it in the inline viewer, or select **Open** for the larger file panel and then select **Download**.



Downloaded videos remain in the conversation. You can find them later through the **Browse files** Files icon in the chat header; choose **Generated** to narrow the list.

## Multiple links and partial results

Bearly handles multiple URLs one at a time. The **Downloading video** step reports the completed and failed counts as it proceeds.

If one of several links fails, the videos that finished successfully remain playable and downloadable. Bearly reports how many were downloaded and how many failed. If every link fails, expand **Get Video** and read **Error** for the immediate problem.

Selecting **Stop** ends the active request. Any video that had already appeared in the conversation remains there.

## What may prevent a download

A direct link to a publicly reachable video page has the best chance of working. A site may reject a download because the video was removed, requires a sign-in, is private, is restricted in your region, or uses a format Bearly cannot save. A site that normally works may also change how it serves videos.

If a download fails:

- Open the URL normally and confirm that the video still plays.
- Copy the address of the individual video page rather than a search, profile, feed, or playlist page.
- Retry with one URL at a time.
- For YouTube, switch to the Bearly desktop app.
- If the source is temporarily unavailable or has changed, try again later.

> **Caution — Download only what you have the right to save**
> Download media only when you own it or have permission or another legal right to save it. Follow the source site's terms and applicable law. Downloading a file through Bearly does not grant permission to reuse or publish it.

---

---
title: "Turn text into speech"
description: "Listen to a Bearly answer or create downloadable spoken audio with the voice and delivery you want."
canonical: "https://bearly.ai/docs/features/text-to-speech/"
markdown: "https://bearly.ai/docs/features/text-to-speech.md"
---

# Turn text into speech

HTML: [https://bearly.ai/docs/features/text-to-speech/](https://bearly.ai/docs/features/text-to-speech/)

> Listen to a Bearly answer or create downloadable spoken audio with the voice and delivery you want.

Bearly can read an answer aloud or turn text into an audio file. Use **Read aloud** when you simply want to listen. Use **Text to Speech** when you want a downloadable recording or need to direct the voice, pace, tone, accent, or emotion.

## Listen to an answer

After Bearly returns a text answer, select **Read aloud** beneath it. Select the same control again to stop playback.

This is the quickest way to listen while proofreading, reviewing a long response, or working away from the screen. If playback does not begin, check that the device is not muted and that the browser or operating system allows audio from Bearly.

## Create an audio file



1. **Choose Text to Speech**

   In a chat, type `/tts`, then select **Text to Speech** from the command menu.

2. **Add the words and direction**

   Paste or type the text you want spoken. If the delivery matters, describe it plainly in the same message.

   ```text copy
   Read this in a measured, warm voice. Pause slightly between the two paragraphs and keep the ending understated:

   [Paste the text here]
   ```

3. **Send the message**

   Bearly creates an audio player in the conversation. Play the recording there, or use its download control to save an MP3.



The generated audio remains with the chat, which makes it easy to compare versions. To change the performance, send the text again with more specific direction—for example, “slower,” “less formal,” or “with a quiet, reassuring delivery.”

> **Tip — Direct the performance, not the technology**
> Describe how the result should sound to a listener. A short note about pace, mood, emphasis, or accent is usually more useful than a long technical specification.


## If the audio is not right

- If a name or unusual term is mispronounced, spell it phonetically in the text and generate another version.
- If the delivery feels exaggerated, simplify the direction to one or two qualities.
- If generation fails, shorten a very long passage and try again after checking the connection.
- If an older recording says it is no longer available, generate it again from the original text. Download important finished recordings rather than relying on an old chat as permanent audio storage.

---

---
title: "Transcribe audio and video"
description: "Turn recordings into searchable text, follow along with playback, label speakers, and download subtitles."
canonical: "https://bearly.ai/docs/features/transcripts/"
markdown: "https://bearly.ai/docs/features/transcripts.md"
---

# Transcribe audio and video

HTML: [https://bearly.ai/docs/features/transcripts/](https://bearly.ai/docs/features/transcripts/)

> Turn recordings into searchable text, follow along with playback, label speakers, and download subtitles.

**Transcribe** turns the spoken content of an audio or video file into text. The result stays connected to the recording, so you can follow along, jump to a moment, download subtitles, or continue asking questions in the same chat.

## Create a transcript



1. **Choose Transcribe**

   Type `/transcribe` in a chat, then select **Transcribe** from the command menu.

2. **Attach the recording**

   Select the plus icon beside the message box, choose **Attach files**, and add one or more audio or video files. Wait for the attachments to finish loading.

3. **Send the message**

   Bearly transcribes every attached recording. You can also add a follow-up request in the same message, such as `Transcribe these interviews, then list the decisions and open questions.`



Audio and video files can be as large as 400 MB each. Whether a particular format plays or uploads can vary by device, so use a common audio or video format if a file is rejected.

## Review the result

The transcript appears below the media player.

- **Captions** follows the current point in the recording and is useful for watching or listening along.
- **Table** shows timestamped passages. Select a passage to jump to that point in the recording.
- **Density** changes how much text Bearly groups into each passage. It does not change the words in the transcript.
- If Bearly identifies different speakers, edit the **Speaker** names shown beneath the transcript. The names are saved when you leave each field.
- When several recordings were submitted together, use the previous and next controls to move among their transcripts.
- Select **Download subtitles** to save the current transcript as an SRT subtitle file.

Bearly also attaches a text copy of a completed transcript to the conversation. This lets you ask for a summary, quotations, action items, a cleaned-up interview, or answers grounded in the recording.

You do not need to choose a language before sending the file. Bearly identifies the spoken language automatically and uses its broader multilingual transcription path when the recording is outside the highest-accuracy language set. For a recording that mixes languages, review the transitions carefully; short phrases and heavy overlap can still be assigned to the wrong language.

> **Caution**
> Names, numbers, accents, and overlapping voices can be transcribed incorrectly. Check important passages against the recording before publishing or relying on exact wording.


## If a file cannot be transcribed

- Confirm that it is an audio or video file and that the attachment finished loading.
- If the file is larger than 400 MB, export a smaller copy or divide it into parts.
- If some files succeed and others fail, keep the completed results and submit the failed recordings again separately.
- If the transcript is empty or inaccurate, check that the recording contains audible speech. A cleaner recording or less background noise may improve another attempt.
- If an older transcript says it is no longer available, attach the original recording and run **Transcribe** again.

---

---
title: "Run prompts on a schedule with Routines"
description: "Create a routine that runs a saved prompt on a schedule, keeps every run in its own chat, and can email or text you the result."
canonical: "https://bearly.ai/docs/features/routines/"
markdown: "https://bearly.ai/docs/features/routines.md"
---

# Run prompts on a schedule with Routines

HTML: [https://bearly.ai/docs/features/routines/](https://bearly.ai/docs/features/routines/)

> Create a routine that runs a saved prompt on a schedule, keeps every run in its own chat, and can email or text you the result.

A routine is a prompt Bearly runs for you on a schedule — a weekday news brief, a Friday roundup of mentions, a month-end recap. Each routine gets its own chat. Every run lands there marked unread, and you can also have the result emailed or texted to you.

You can create and edit routines anywhere, but they run in the Bearly desktop app, on the one device you install them on, while the app is open.

## Create a routine



1. **Open Routines**

   Select **Routines** in the sidebar, or type `/routine` in any chat. You can also ask in a conversation: when a chat is already doing the task, "do this every weekday at 8" turns it into a routine, with the same options as the editor.

2. **Describe it**

   Type what you want and when into the box — for example, "AI news brief, weekday mornings" — and select **Draft routine**. Bearly turns your description into a draft. **Start from scratch** opens a blank editor instead.

3. **Review the draft**

   Check the **Name**, **Instructions**, **Model**, and **Schedule**. Write the instructions so the first run has everything it needs, then choose the model you want Bearly to use for every run. The model list follows your plan and your organization's model policy. Schedules can be daily, weekdays, weekends, chosen days of the week, or a day of the month, at an exact time in your timezone. The editor shows the next few run times so you can confirm it reads your intent correctly.

4. **Select Create routine**

   The routine gets its own chat, named after it. If you created it on desktop, it starts running there; if you created it on web or mobile, install it from the desktop app when you're at that machine.



You can keep up to 10 active routines, and runs must be at least 30 minutes apart.

## Give runs tools with plugins

In the editor, **Tools** → **Tools and skills** controls what each run can use beyond web search, X search, and the files in the routine's chat:

- **Connectors** — let runs read your connected apps, like a calendar, or a [server you added yourself](https://bearly.ai/docs/features/connectors.md). Each connector has a **Tools** setting: **All** or **Read-only**. Google Workspace is not available in Routines. A chat approval does not grant unattended Routine access. Connectors sign in per device, so if one isn't connected on the device that runs the routine, the run says so instead of failing silently.
- **Skills** — give runs your installed skills. A skill with scripts also has an **Allow scripts** switch; turning it on lets that skill's scripts run unattended on your machine. Leave it off unless you trust the skill to run without you watching.
- **Deep Research** — each run does full paid research and emails you the PDF report. It's billed per run at the **Standard** or **Ultra** tier; the editor shows each tier's current per-run price, since pricing changes over time. Do the math before you put research on a daily schedule.

## Choose the routine's permission mode

Routines start in **Standard**, which uses the unattended permissions you have already enabled for that routine. **Full access** skips additional action approvals for its selected tools. It does not turn on a tool, connect an account, or expand what Routines can do. Your chat's permission mode does not carry over.

Full access has limited availability and requires a supported account and desktop app version.



1. Create the routine, or select **Save** after editing it.
2. On the desktop device shown in **Run location**, open the routine and find **Tools** → **Permission mode**.
3. Select **Enable full access**, review the selected tools and warning, then select **Enable full access** in the dialog. Choose **Cancel** to keep the current mode.



The routine shows **Full access** in its editor and in the list. Enabling it stops any current run and leaves the schedule unchanged. For a managed organization account, an organization administrator must allow Full access first. The routine's selected tools, connector accounts and tool restrictions still apply, including a connector's **Read-only** setting.

Unlike a chat, a routine keeps this permission across app restarts and time away, so scheduled runs can work unattended. Bearly checks authorization for every run. The desktop app must still be open, and its required tools and connections must be available.

Saving changes to the instructions, schedule, model, delivery settings or selected tools stops the current run and requires a new review. Changing a selected connector's sign-in or an installed skill can also require review. The routine shows **Full access needs review** and cannot continue in Full access until you select **Review full access** and approve the current setup on its running device. If its schedule has paused, select **Resume** after reviewing. Unsaved edits leave the saved routine unchanged. Copies and routines moved to another device start in Standard.

Select **Use Standard** to turn off Full access and stop the current run. Existing per-tool unattended permissions remain, including local or browser access you enabled separately. Standard routines can still act through those tools without asking; turn off the corresponding tool settings to remove that access. **Stop run** ends just the current run and keeps its schedule and permission mode. A change made from another device takes effect when it syncs to the running device. Web and mobile can manage these settings, but cannot enable Full access or run the routine.

If **Permission mode** cannot load, check your connection and select **Retry**. Follow any message to update Bearly or enable an available tool. If your organization no longer allows Full access, choose **Use Standard** or ask an organization administrator to restore eligibility. Restored eligibility still requires a new review; it does not resume the routine automatically.

Full access does not add screen or computer control, Google Workspace, or Chrome DevTools to Routines. Browser access stays limited to the routine's own tabs and to the sites and switches you set under [Let a routine use a browser](#let-a-routine-use-a-browser).

## Allow commands and local file access

On the desktop app, **Tools** → **Unattended local access** lets a routine use Local Exec and selected skill scripts without waiting for approval. Bearly shows a detailed warning and does not turn the setting on until you select **Allow unattended access**. If you ask Bearly to enable Local access while setting up a routine in chat, an approval card with the same risks appears before the routine is saved; select **Allow** to continue or **Deny** to stop the save without enabling it. Asking Bearly in chat to change the instructions of a routine with Local access shows the card again, because the instructions are what you approved.

> **Danger — The routine can change or delete data**
> With unattended local access on, the AI can run shell commands and skill scripts and can read, upload, change, overwrite, or permanently delete files available to your computer account. It acts without asking before each command. AI mistakes or malicious instructions hidden in websites, files, email, or connector data could expose secrets or destroy data.


Leave this off unless the routine needs local commands. Keep current backups, give the routine the smallest practical working folder, and test the saved routine with **Run now** before relying on its schedule. You can turn the setting off at any time; later runs will stop receiving Local Exec.

This setting does not override your computer's permissions or your organization's Bearly policy, and it never lets a scheduled run see your screen or control your computer. Browser access is a separate setting, described next.

**Build on previous runs** under **Results and delivery** is on by default for new routines. Each run sees the chat's earlier runs — useful for "tell me what changed since yesterday" or "don't recommend anything you've already suggested." Turn it off when every run should start clean. Existing routines keep their saved setting.

## Let a routine use a browser

On the desktop app, **Tools** → **Browser access** lets scheduled runs use [Bearly Browser](https://bearly.ai/docs/features/browser.md) on the sites you list, while you're away. Nothing it can do there needs your approval at the time, so the list and the switches below are the limits, with or without Full access. Routines can't use your Chrome.

- **Sites** — the exact sites the routine can open, like `example.com`. `www.` is optional, but every other subdomain is its own site: add `pay.example.com` or a sign-in page on another domain separately. Sites named in the instructions appear under **From the instructions** so you can add them in one click. The routine can't open, read, or act on any other site, and a page it opens can't send it to one: the desktop app stops the request before it goes out. The routine says where it stopped in **Recent runs**.
- **Click, type, and submit** — off by default. When off, the routine only reads pages and follows links. Turn it on when the routine has to click buttons, type, choose options, or submit a form.
- **Download files** — off by default, and available only with **Click, type, and submit** on. Lets the routine save files from these sites into its chat.

When you save, a warning lists everything the routine can do and what's new since you last approved it, and nothing is saved until you select **Allow browser access and save**. Removing a site or turning a switch off saves without asking. If you ask Bearly in chat to set up browser access, an approval card with the same details appears before the routine is saved; select **Allow** to continue or **Deny** to save nothing. The card appears again whenever the access grows, or when Bearly changes the instructions of a routine that has browser access.

The routine uses the site sign-ins in Bearly Browser on the device that runs it, including cookies you [imported from Chrome or Edge](https://bearly.ai/docs/features/browser.md#bring-in-existing-site-sign-ins). The Browser panel stays closed while the routine works, and the routine's tabs close when the run ends. **Enable Browser** and **Allow chats to use Browser** must be on for that device, and the desktop app must be up to date.

> **Danger — The routine acts as you on the sites you allow**
> With **Click, type, and submit** on, a run can send messages, make purchases, and change or delete account data on those sites without asking. Instructions hidden in a web page, email, or document can trick it into actions your instructions didn't ask for, and sites treat its actions as yours.


Allow only the sites the routine needs, leave **Click, type, and submit** off unless it has to change something, and test with **Run now** before relying on the schedule. Sign in ahead of time: a run never clicks or types into password or payment fields, and can't upload files. No run accepts an invalid HTTPS certificate; the routine closes that tab and reports the problem.

If the instructions or the browser access change on a device that can't check them, such as an older version of Bearly, later runs go without a browser until you open the routine and save it to approve them again. Routines that had browser access before sites existed also run without a browser until you pick their sites.

If Bearly Browser isn't available when a run starts, the run goes ahead without it and says so in **Recent runs**. That happens when your organization doesn't allow it, when Browser is off on that device, or when the desktop app needs an update. A Full access run also needs current authorization for its remaining tools; if its permissions need review, it stops instead. You can turn browser access off at any time; later runs won't use a browser.

## Control when you hear about it

Every run lands in the routine's chat either way. The **Notification filter** field sets the bar for actually notifying you — for example, "a competitor ships something new". When a run doesn't meet the bar, there's no unread badge, email, or text; the result just sits quietly in the chat, and **Recent runs** notes why. So that silence never reads as a dead routine, it checks in anyway every 7th quiet run in a row — and when the bar can't be checked, it delivers to be safe.

Two delivery options sit below it. They tell the routine's chat to send the finished result before the run ends, so the send appears in the chat with the rest of the run:

- **Email each result** — sends each delivered result to your account email. Routines only email you, never other addresses.
- **Text each result** — texts a short AI summary of each delivered result to your verified number. If texting is not ready, turning this on opens **Settings** → **Phone & SMS**. Add or re-verify your number there; after verification, the routine's text option turns on. Texts are capped at 3 a day, and a text leaves end-to-end encryption — the switch says so where you turn it on.

If a send fails, open the routine's chat and read the failed **Send Email** or **Send SMS** step. When SMS needs setup, the chat also shows an **Open Phone & SMS** button. The run result stays in the chat, and you can ask Bearly to send it again after fixing the account or phone setting. Check whether the first message arrived before retrying so you do not create a duplicate.

## Where routines run

Routines run on one device: the **Run location** row in the editor shows which. **Install on this device** claims a routine created elsewhere; **Move to this device** takes one over from another machine. A routine that isn't installed anywhere doesn't run.

The desktop app has to be open. If it was closed or asleep over a scheduled time, the routine catches up when the app is back: it runs once for the most recent missed slot, and older missed slots are recorded as skipped in **Recent runs** rather than run late in a burst.

**Run now** in the editor runs the last saved version immediately on the current device — useful to test a new prompt without waiting for the schedule. A Full access routine must run on its installed desktop device.

## When a routine fails

**Recent runs** in the editor shows each run's outcome, including skips and why. If a routine fails twice in a row it pauses itself and shows a Bearly notification. Fix the cause — often a connector that needs reconnecting on that device — then select **Resume**. Individual runs are also stopped if they take longer than about 25 minutes.

A run that uses Bearly Browser stops if you reset Browser tabs, delete Browser data, or turn off Browser while it works; **Recent runs** says so. If you take over one of the routine's Browser tabs, the run can't use that tab and reports what it finished.

If you delete a routine's chat, its next run creates a new chat with the routine's name. Standard routines continue there; a Full access routine needs a new permission review before continuing in the replacement chat. The deleted chat and its contents are not restored. If Bearly cannot create the replacement, the routine pauses and **Recent runs** shows that the chat was missing.

Pausing a routine stops its schedule and current run but keeps its settings and history. **Stop run** stops current work without pausing future runs. Deleting a routine stops it; its chat and past runs stay in your history. Changes made on another device take effect on the running device when they sync. None of these controls undo actions already completed in another system.

## What scheduled runs can and can't do

A scheduled run works unattended, so it's deliberately more limited than you working in a chat. It can search the web and X, read and search the files in its chat (including PDFs and saved transcripts), write files, use the connectors and skills you selected, and run paid Deep Research if you enabled it. By default, it does not run general commands on your machine; skill scripts run only for skills where you turned on **Allow scripts**. Turning on **Allow unattended local tools** also gives the routine Local Exec and permits selected skill scripts without individual approval.

Scheduled runs never see your screen or control your computer. They use a browser only when you turn on **Browser access**, and then only Bearly Browser, on the sites you list.

For anything that needs your hands or eyes, open the routine's chat and continue the conversation there. It's a normal Bearly chat with everything you have anywhere else, and the routine's full run history as context.

Administrators of managed organizations can turn Routines off from **Organizations** → organization name → **Policies** → policy name → **Code & Automation** → **Routines**. **Browser in Routines** in the same section controls browser access; a routine also needs **Browser** allowed.

For organization eligibility and the difference from interactive chats, see [Choose how Bearly uses external tools](https://bearly.ai/docs/settings/agent-permissions.md).

---

---
title: "Record and transcribe meetings"
description: "Record online meetings, in-person conversations, and voice notes in Bearly, with a live transcript, speaker names, summaries, and a chat that starts from the transcript."
canonical: "https://bearly.ai/docs/features/meetings/"
markdown: "https://bearly.ai/docs/features/meetings.md"
---

# Record and transcribe meetings

HTML: [https://bearly.ai/docs/features/meetings/](https://bearly.ai/docs/features/meetings/)

> Record online meetings, in-person conversations, and voice notes in Bearly, with a live transcript, speaker names, summaries, and a chat that starts from the transcript.

**Meetings and Recordings** records a meeting, a call, or a voice note and transcribes it as people speak. When you stop, the audio and transcript are saved to your account, and you can name speakers, get a summary, copy or download the transcript, or start a chat with it.

<ThemeScreenshot
  lightSrc="/images/docs/features/meetings/meeting-notes-light.png"
  darkSrc="/images/docs/features/meetings/meeting-notes-dark.png"
  alt="Bearly Meetings with the conversation list open and a Product onboarding review recording showing its summary, named speakers, and transcript."
/>

## Record a meeting



1. **Open Meetings and Recordings**

   Select **Meetings and Recordings** in the sidebar.

2. **Choose what you're recording**

   - **Online meeting** records the call and your voice together — Zoom, Teams, or anything else playing on your computer. This option needs the Bearly desktop app; on web and mobile the card offers **Get desktop app** instead.
   - **In person or voice note** records the room through the device's microphone and works on every platform. Audio you hear in headphones isn't included.

3. **Select Start recording**

   Allow microphone access if Bearly asks. On a Mac, recording computer audio also uses the system's Screen Recording permission; if it's missing, Bearly shows a notice with an **Open Screen Recording settings** button. If computer audio can't start, the recording continues from the microphone and says so.

4. **Select Stop recording when you're done**

   The transcript you watched live is saved and shown immediately. The audio uploads to your account in the background, and Bearly reviews the full recording for more accurate text and speaker labels. The final transcript replaces the live one when it's ready — you don't need to stay on the page.



While recording, the live transcript appears in the recording's entry at the top of **Recordings**. You can work elsewhere in Bearly: with the sidebar open, a compact **Recording** card shows the latest transcript line, and with the sidebar closed a floating stop control stays visible. Audio is saved continuously on the device as you record.

The live transcript supports English, Spanish, French, German, Italian, Portuguese, Arabic, Danish, Dutch, Hebrew, Hindi, Japanese, Mandarin Chinese, Vietnamese, Finnish, Norwegian, Swedish, and Turkish. It can follow speakers who switch between supported languages in the same conversation. After you stop, Bearly checks the complete recording and can use broader language coverage for the final transcript when needed.

A recording can run up to **1 hr 15 min**. At the limit, Bearly stops the recording and keeps everything captured up to that point.

> **Caution**
> Only record people with their permission. Recording laws vary by place; you are responsible for getting consent.


## Work with a recording

Select a recording in the **Recordings** list. Bearly proposes a title from the first minutes of the conversation; edit the title any time, and a title you type is never replaced.

- **Speakers** — after you stop, Bearly automatically reviews the full recording to tell voices apart more accurately. When more than one voice is identified, a form lists each detected speaker. Type a name once and it applies to the live view, the saved transcript, and exports. A single-voice recording reads as a voice memo, without speaker labels.
- **Improve speaker labels** — older recordings that have not had the automatic speaker review may show this button. It runs the same full-recording review, and names you already typed carry over.
- **Create summary** — writes a quick summary with key points, decisions, action items, and open questions, and saves it with the recording. **Refresh** rewrites it; **Copy** puts it on the clipboard.
- **Chat with transcript** — opens a new chat with the transcript attached as a file, ready in the composer. Nothing is sent, and nothing is spent, until you send a message.
- **Copy** — opens **Copy transcript**, where you choose **Text only**, **Speaker names**, or **Names and timestamps** (for a voice memo, **Text only** or **Timestamps**), then copy the result or select **Download .txt**.
- The audio player plays the original recording, and you can download the audio file.

Completed recordings and transcripts are saved to your account, so they're available when you sign in on another device.

To remove a recording, select the **Delete recording** trash button. Bearly permanently deletes the audio, transcript, speaker names, and summary; this cannot be undone.

> **Caution**
> Names, numbers, accents, and overlapping voices can be transcribed incorrectly. Check important passages against the audio before relying on exact wording.


## If something interrupts a recording

Recordings are built to survive trouble:

- If the connection drops or your usage limit is reached mid-meeting, the live transcript pauses and the audio keeps recording. The recording says **Live transcript paused at the account limit. Audio recording continues.** when a limit is the cause.
- If the final transcript fails, the recording is marked **Needs attention**; select **Retry** to run it again. The audio stays saved either way.
- If an upload can't finish, the recording stays on the device that made it, marked **Saved on device**, with **Retry upload** and **Discard** in its detail pane. **Discard** permanently removes the local audio and transcript.
- If microphone access is denied on desktop, the error includes an **Open microphone settings** button that takes you to the right system pane.

On the desktop app, an in-progress recording keeps going while you navigate around Bearly, and the app finishes saving before it quits. In a browser, keep the recording tab open: audio already saved survives a refresh, but the live capture stops.

<span id="for-team-administrators"></span>

## For organization administrators

Administrators of managed organizations can turn Meetings off from **Organizations** → organization name → **Policies** → policy name → **Collaboration** → **Meetings**.

Already have a recording as a file? [Transcribe audio and video](https://bearly.ai/docs/features/transcripts.md) turns uploaded audio and video into transcripts in a chat.

---

---
title: "Dictate instead of typing"
description: "Turn on dictation, speak into the message box with a click or a held key, and choose what happens to the finished transcript."
canonical: "https://bearly.ai/docs/features/dictation/"
markdown: "https://bearly.ai/docs/features/dictation.md"
---

# Dictate instead of typing

HTML: [https://bearly.ai/docs/features/dictation/](https://bearly.ai/docs/features/dictation/)

> Turn on dictation, speak into the message box with a click or a held key, and choose what happens to the finished transcript.

Dictation turns speech into text in the message box. You speak, Bearly transcribes, and the finished text lands in your draft so you can read it, edit it, and send it like anything else you typed.

Dictation is off until you turn it on. Nothing is recorded before that.

> **Note — Dictation and voice conversations are different**
> Dictation produces text for you to send. A [voice conversation](https://bearly.ai/docs/features/voice-conversations.md) is a spoken call where Bearly answers aloud. The dictation control is the sound-wave icon in the top-right corner of the message box; the voice control is the microphone icon below it.


## Turn on dictation



1. **Select the dictation button**

   It is the sound-wave icon in the top-right corner of the message box, labeled **Start dictation**. While dictation is off, selecting it opens **Settings** → **Chats** → **Dictation** instead of recording.

2. **Turn on Dictation**

   Use the switch beside the **Dictation** heading. The controls below it become available.

3. **Close settings and speak**

   Select the dictation button again. Allow microphone access when your browser or operating system asks. The button turns into a stop square, a status panel appears at the top of the window, and you hear a short rising tone.



Your dictation settings follow your Bearly account, so turning it on once covers your other devices.

## Start and stop recording

Select the dictation button to start and select it again to stop. You can also hold a key while you speak: keep <kbd>Right Shift</kbd> down, talk, and let go. A quick tap does nothing — hold the key for the recording to begin.

While you are recording, the status panel shows **Listening…** and the words as they are recognized. That text is a live preview; the version that reaches your draft is the finished transcript.

The keyboard shortcut works while the Bearly window is focused and needs a physical keyboard. On a phone or tablet, use the dictation button.

If you press any other key during a held recording, Bearly cancels it and inserts nothing. That is what keeps ordinary shifted typing from producing a transcript.

## Change the shortcut

Open **Settings** → **Chats** → **Dictation** and select the key shown beside **Keyboard shortcut**. Press the combination you want — up to six keys — and release. Bearly stores exactly what you pressed, so <kbd>Left Shift</kbd> and <kbd>Right Shift</kbd> stay distinct. The switch beside it turns the shortcut off if you would rather only use the button.

## Choose where the text goes

**After dictation** controls what happens to the finished transcript:

- **Fill the composer** (the default) adds the text to your draft. If the draft already has text, dictation is appended to it.
- **Send immediately** sends the transcript as a message as soon as it is ready.

If you finish a recording without saying anything, nothing is inserted.

## Clean up the transcript

**Transcript cleanup** is off by default. Turn it on and Bearly runs the finished transcript through a fast model that fixes punctuation and capitalization, and removes filler words and false starts. Its controls stay visible while it is off so you can see what it adds.

You can pick the **Model** and edit the **Instructions** that model follows. Cleanup runs on the text only — the original audio is not sent to it. Cleanup adds a moment before the text appears, and if it fails, Bearly uses the raw transcript rather than losing your words.

## Teach it your words

Under **Words and names**:

- **Vocabulary** is for names, acronyms, and technical terms you use — one per line or separated by commas. It guides cleanup toward the right spellings.
- **Replacements** are exact substitutions applied to every transcript, one `spoken → written` pair per line. `bear lee → Bearly` rewrites the spoken form wherever it appears as a whole word.

Replacements always apply. Vocabulary only affects the result when cleanup is on.

## Limits and availability

A recording stops on its own after 10 minutes and the transcript so far is delivered normally. For longer dictation, record in passes.

Dictation works in the Bearly web app, the desktop app, and the iOS and Android apps, wherever a microphone is available and permitted. On a managed team, an administrator can turn dictation off; the button then reads **Dictation is disabled by your team**, and the settings pane says your team has disabled audio transcription.

Live audio is sent through Bearly for transcription. Bearly does not store the recording.

## If dictation does not work

| What you see | What to do |
| --- | --- |
| Selecting the button opens settings instead of recording | Dictation is still off. Turn on the switch beside the **Dictation** heading. |
| **Dictation is disabled by your team** | Your team policy does not allow it. Ask a team administrator. |
| **Microphone access failed. Check your browser or system permission.** | Allow the microphone for Bearly in your browser's site permissions or your operating system's privacy settings, then try again. |
| **Transcription is unavailable. Try again.** | Bearly could not open the transcription connection. Check your network and start another recording. |
| **Dictation could not finish. Try again.** | The recording was interrupted and no text is inserted. Start a new recording; nothing from the interrupted one is kept. |
| The held key does nothing | Make sure the Bearly window is focused, that the shortcut switch is on, and that you are holding the key rather than tapping it. |
| Cleanup seems to do nothing | Confirm **Transcript cleanup** is on and that a **Model** is selected. Bearly falls back to the raw transcript when cleanup cannot run. |
| A replacement never applies | Check that the line has a `→` or `=>` between the spoken and written forms, and that the spoken form is a whole word in the transcript. |

---

---
title: "Talk with Bearly by voice"
description: "Start a live voice conversation, manage the microphone and call, and continue from the saved transcript."
canonical: "https://bearly.ai/docs/features/voice-conversations/"
markdown: "https://bearly.ai/docs/features/voice-conversations.md"
---

# Talk with Bearly by voice

HTML: [https://bearly.ai/docs/features/voice-conversations/](https://bearly.ai/docs/features/voice-conversations/)

> Start a live voice conversation, manage the microphone and call, and continue from the saved transcript.

Voice conversations let you speak with Bearly in real time inside an existing chat. Bearly can use the conversation's earlier context, answer aloud, and keep a transcript so you can continue the work in writing afterward.

Voice is available wherever the microphone control appears in Bearly. It requires microphone access and a secure connection. It is temporarily unavailable in Firefox; use Chrome or the Bearly desktop app instead. A managed-team policy can also hide voice conversations.

To speak your message instead of holding a conversation, use [dictation](https://bearly.ai/docs/features/dictation.md).

> **Note — The chat keeps a transcript, not a recording**
> Bearly saves the final transcript and call status with the conversation. It does not save the microphone audio as a chat recording.


## Start a voice conversation



1. **Open the chat you want to continue**

   Voice starts inside the current chat. If you want Bearly to draw on a document, earlier answer, or other context, open that conversation before starting the call.

2. **Select Start voice call**

   Select the microphone button labeled **Start voice call** beside the message box. When your browser or operating system asks, allow Bearly to use the microphone.

3. **Wait for Listening**

   The message box becomes a call bar. Its status moves through **Connecting**, **Listening**, **Thinking**, and **Speaking**, so you can tell whose turn it is. The active call also appears as a microphone indicator on that chat in the sidebar.

4. **Speak naturally**

   Ask a question, develop an idea, or continue the task already in the chat. Bearly answers aloud and can use the capabilities available to that conversation when your request needs them.

   You can start speaking while Bearly is answering; your voice interrupts the reply so the conversation can change direction without waiting. In a noisy room, use headphones and keep the microphone close. If nearby speech starts unwanted turns, mute the microphone until you are ready to speak.

5. **End the call**

   Select **End call** in the call bar or on the voice card in the conversation. The finished card shows **Call ended**, the duration, and **Show transcript**.



You can follow the call with an ordinary message such as “Turn that discussion into a one-page brief.” The saved transcript gives Bearly the text of the voice conversation; you do not need to repeat it.

## Use the call controls

- **Microphone**: When more than one input is available, choose the microphone by selecting its name in the call bar. Bearly remembers the selected device on that device.
- **Mute microphone**: Stops sending microphone input without ending the call. Select **Unmute microphone** when you are ready to continue.
- **End call**: Ends the session and returns the normal message box.
- **Show transcript**: Expands the completed conversation as alternating **You** and **Bearly** turns.

Keep Bearly in the foreground during a mobile voice conversation. Mobile operating systems can pause live microphone and network activity when the app moves to the background.

## Choose the voice model

Calls use **GPT-Live 1** unless you pick another model. On a paid plan, open your account menu, select **App Settings**, and choose **Chats** → **Voice**. Under **Model**, pick **GPT-Live 1**, **Gemini 3.8 Live**, or **Gemini 3.8 Live Extended Thinking**. The choice saves right away, syncs to your other devices, and applies to your next call.

During a call, the call bar and the call card name the model in use next to the timer. If your chosen model was unavailable and the call started on another one, the call bar adds **fallback** to that name.

On a Gemini model, the voice you hear changes with the model, and permission requests appear as cards in the chat instead of spoken questions. Everything else on this page works the same way.

The Free plan runs calls on a smaller model, and the **Model** control stays off with a note that choosing a voice model needs a paid plan.

## Usage and session limits

Voice conversations count toward your Bearly usage. The model and per-session allowance depend on your plan. If a call reaches its allowance, Bearly shows **Voice limit reached** and saves the transcript completed so far. Start another call when you are ready to continue.

A call ends after 30 minutes, or sooner if it stays quiet for a short period. Select **Start voice call** again to begin a fresh session in the same chat.

## If voice is unavailable

| What you see | What to do |
| --- | --- |
| The microphone control is missing | Check for a Bearly update. If this is a managed account, ask a team administrator whether voice is allowed by your policy. |
| **Voice is temporarily unavailable in Firefox** | Open the chat in Chrome or the Bearly desktop app. |
| **Microphone access is off** | In the desktop app, select **Settings** from the voice notice, enable microphone access in the operating system, and try again. On the web, allow the microphone in the site's browser permissions, then reload the chat. |
| **No microphone is available** | Connect or enable a microphone, confirm that another app is not preventing access, and select **Start voice call** again. |
| The wrong microphone is active | Open the microphone selector in the call bar and choose another input. If that device is no longer connected, Bearly falls back to an available microphone. |
| **Voice connection closed** or **Voice connection stalled** | Start another call. If the call ended before 30 minutes, check the network connection first. Any completed transcript already shown in the chat remains available. |
| **Voice limit reached** | Dismiss the notice or wait for it to close, then select **Start voice call** for a new session. |

---

---
title: "Add workflows with Skills"
description: "Install, manage, and use reusable Bearly workflows from the Skill Store or a trusted source."
canonical: "https://bearly.ai/docs/features/skills/"
markdown: "https://bearly.ai/docs/features/skills.md"
---

# Add workflows with Skills

HTML: [https://bearly.ai/docs/features/skills/](https://bearly.ai/docs/features/skills/)

> Install, manage, and use reusable Bearly workflows from the Skill Store or a trusted source.

Skills are reusable packages of instructions and optional supporting files. They give Bearly a consistent method for a particular kind of work, such as reviewing a document, preparing a deliverable, or following a team's process.

Installing and managing Skills requires the Bearly desktop app for macOS, Windows, or Linux. A conversation that used a Skill remains readable on the web or on mobile, but return to the desktop app when the Skill needs its installed files or needs to run a local action.

## Install a Skill from the Skill Store



1. **Open Skills from a chat**

   In the Bearly desktop app, select the plus icon beside the message box. Under **Settings**, select **Skills**. Bearly opens **Settings** → **Skills** at the **Skill Store**.

2. **Browse a repository**

   Under **Skill repositories**, select a repository to expand it. Bearly reads the repository when you open it and then lists the Skills it contains.

3. **Review the source before installing**

   Select **Install** beside the Skill you want. Read the **Install third-party skill?** notice, including the source location and its warning about instructions and scripts.

4. **Confirm the installation**

   Select **I understand the risks** only when you trust the source. The Skill then appears under **Installed** with an **Enabled** status.



If your organization controls who can add Skills, the store may offer **Request** instead of **Install**. You can still browse available repositories and pause Skills that are already installed.

> **Caution — A Skill can contain actions, not just advice**
> Third-party Skills can include instructions and files that Bearly may run on your computer. A malicious or mistaken Skill could change or delete data. Review the source, consider which files it can reach, and do not install a Skill merely because its name sounds useful.


## Use a Skill in a chat



1. Open a new or existing chat in the desktop app.
2. Type `/skill`, then select **Use _skill name_** from the **Skills** group. The chat shows a **Skill _skill name_** control beside the composer.
3. Describe the work you want done. Include the source material, desired result, and any boundaries that matter.
4. Send the message. Bearly opens the selected Skill and follows its instructions while answering.



An enabled Skill may also be used when your request clearly matches its purpose. Selecting it with `/skill` is the clearest way to tell Bearly which workflow you want.

In Standard, if the Skill needs to run a local file, Bearly shows a **Run script** permission request. Open **Details** to review the command, reason, and other information. Choose **Once**, **This response**, **This session**, or **This chat**, then select the Allow button, or select **Deny** to keep it from running. A longer approval covers scripts from that Skill installation. Chat approvals stay on the current device; revoke them under **Saved approvals** in **Permissions** near the chat input. Denying the action does not prevent you from continuing the chat with a different approach.

[Full access](https://bearly.ai/docs/settings/agent-permissions.md) lets the chat run scripts from enabled Skills without asking for tool approval. Skills you deliberately enable later are included; organization restrictions still apply.

## Pause, enable, or remove a Skill

Return to the plus icon → **Skills**, then use the controls under **Installed**:

- **Pause** keeps the Skill installed but removes it from the chat's available Skills. Select **Enable** when you want to use it again.
- **Uninstall** removes an installed Skill from this desktop.
- **Unlink** removes Bearly's connection to a linked folder without deleting the original folder.

A Skill marked **Managed by your team** comes from a team policy. You can pause or enable it, but a team administrator must change or remove the policy assignment.

## Install from a trusted source

The Skill Store is the simplest starting point. If a trusted publisher gives you installation details, open **Install from source**:

- Under **Git**, enter the repository address and select **Add Repository**. Keep the supplied folder and version details, or leave the defaults when the publisher does not specify them. Expand the new repository under **Skill repositories** to install an individual Skill.
- Under **URL ZIP**, enter the HTTPS address and its **SHA-256** verification code, then select **Install**. The publisher should provide both values.
- Under **Folder**, select **Import Folder** to copy a Skill into Bearly, or **Link Folder** to keep using it from its original location. You can choose one Skill folder or a parent folder that contains several Skills.

Each Skill folder must contain the file that identifies it as a Skill. If Bearly reports **No SKILL.md files found in this repository**, check the publisher's instructions or choose the correct folder.

> **Note**
> Public GitHub repositories can still be added when Bearly reports that Git is unavailable. Private repositories require Git to be available on the computer.


## Use Skills in a project or managed team

Installed Skills belong to the current desktop. A project or team policy can also decide which of those Skills are available in particular chats.

- In a project, open **Settings** and find **Skills**. Project owners and writers can add, pause, or remove shared Skills; other collaborators see **View only**.
- On a managed team, an administrator can supply Skills or limit whether members may add, use, or run them. See [Set team policies](https://bearly.ai/docs/teams/policies.md).

A Skill can therefore be installed and enabled on your computer but absent from one chat because that project's settings or your team's policy does not include it.

## If a Skill is unavailable

| What you see | What to do |
| --- | --- |
| **Use Skills in the desktop app** | Open the same account in Bearly for macOS, Windows, or Linux. |
| **Update Bearly to use Skills** | Install the current desktop release, reopen Bearly, and try again. |
| The Skill is missing after `/skill` | Open **Settings** → **Skills**, confirm it is **Enabled**, then reopen the chat. Also check whether the chat's project or team policy limits Skills. |
| A repository will not load | Check your connection, close and reopen **Skills**, then expand the repository again. |
| A linked Skill stopped working | Make sure its original folder still exists at the same location. If it moved, **Unlink** it and use **Link Folder** again. |
| A local action was denied | Send a new message with a safer plan, or retry and select **Allow** only after the **Run script** details match your intent. |

If all you need is a reusable instruction with a few fill-in fields, [create a Prompt Template](https://bearly.ai/docs/features/prompt-templates.md) instead. Prompt Templates do not install files or run local actions.

---

---
title: "Create reusable Prompt Templates"
description: "Turn a recurring instruction into a reusable form that you can fill out and submit in any Bearly chat."
canonical: "https://bearly.ai/docs/features/prompt-templates/"
markdown: "https://bearly.ai/docs/features/prompt-templates.md"
---

# Create reusable Prompt Templates

HTML: [https://bearly.ai/docs/features/prompt-templates/](https://bearly.ai/docs/features/prompt-templates/)

> Turn a recurring instruction into a reusable form that you can fill out and submit in any Bearly chat.

Prompt Templates save instructions you use repeatedly. A template can be a fixed prompt, or it can present a small form for details that change each time—such as source text, audience, format, or an attached file.

Unlike a Skill, a Prompt Template does not add a workflow package or local actions. It turns your saved instruction and the values you enter into an ordinary message for the current chat.

## Create a Prompt Template



1. **Open Prompt Templates**

   Select **Prompt Templates** in the Bearly sidebar.

2. **Describe the recurring task**

   If this is your first template, select **Create Prompt Template**. Otherwise, select **New Prompt**.

   In **AI Prompt Builder**, answer **What would you like to create?** Describe the desired result and the information that should change each time, or choose one of the **Popular Ideas** as a starting point. Then select **Create Prompt**.

3. **Review the generated template**

   Bearly opens the new template. Under **Basic Information**, refine its **Template Name** and **Description**. Under **Template Design**, review **Your Template** and the **Live Preview**.

4. **Add or adjust fill-in fields**

   In **Your Template**, wrap a field name in double braces wherever a value should be supplied later. For example:

   ```text copy
   Summarize {{ meeting notes }} for {{ audience }}. End with {{ next step format }}.
   ```

   Bearly creates one form field for each unique name. A template can also contain no fields when the instruction should be exactly the same every time.

5. **Configure each field**

   Under **Field Configuration**, open a field and choose what the person using the template will see:

   - **Helper Text** explains what belongs in the field.
   - **Short Answer** provides a compact input; **Paragraph** provides room for longer material.
   - **Allow File Uploads** adds an attachment control to that field.



Bearly saves changes as you edit; there is no separate save button. The **Live Preview** shows how the form will appear in chat.

> **Tip**
> Name fields by the information they need—`{{ source text }}`, `{{ audience }}`, or `{{ output format }}`—rather than vague names such as `{{ input }}`. Clear fields make a template easier to reuse months later.


## Use a template in chat



1. Select **Chat** in the sidebar and open the conversation where you want to use the template.
2. Type `/`, then find your template under **Prompt Template** and select its name.
3. Fill in the fields. If a field allows files, select its paperclip icon and attach the relevant file. A fixed template with no fields shows the complete instruction for review.
4. Select **Submit**. You can also press <kbd>Shift</kbd> + <kbd>Enter</kbd> while the form is open.



Bearly replaces each field in the saved instruction with what you entered and submits the result to the current chat. Fill every field you expect the answer to use: a field left blank is sent as `N/A`.

The filled template remains visible in the conversation. If you change a value there, **Reset** discards those edits and **Submit** sends the revised version for a new answer.

## Edit or delete a template

Open **Prompt Templates**, then select a template under **My Prompts**. Edit its name, description, instruction, or fields directly; changes are saved as you type.

To delete it, select the three-dot menu beside its name under **My Prompts**, choose **Delete**, then confirm **Delete**. Bearly removes it from the Prompt Templates page and from the chat command menu; the app does not show an undo action.

Deleting a template does not remove messages that were already submitted with it. Those remain part of their conversations.

## If a template does not work as expected

| What you see | What to do |
| --- | --- |
| The template is missing from the `/` menu | Make sure **Your Template** is not blank. Wait a moment for the edit to save, then reopen the chat and search for the template's **Template Name** again. |
| **Submit** does nothing | When a template has fields, enter at least one value or finish attaching a file before submitting. |
| An attachment shows a warning icon | Remove the file from the field, check your connection, and attach it again. |
| **Failed to create prompt. Please try again.** | Keep your description, check your connection, and select **Create Prompt** again. |
| The command menu still shows an old name | Reopen the chat after renaming the template so the command list refreshes. |

For a reusable workflow that includes its own supporting files or approval-gated local actions, see [Add workflows with Skills](https://bearly.ai/docs/features/skills.md).

---

---
title: "Compare answers with Council of Experts"
description: "Ask several AI models the same question, inspect their individual answers, and use Bearly's synthesis."
canonical: "https://bearly.ai/docs/features/council-of-experts/"
markdown: "https://bearly.ai/docs/features/council-of-experts.md"
---

# Compare answers with Council of Experts

HTML: [https://bearly.ai/docs/features/council-of-experts/](https://bearly.ai/docs/features/council-of-experts/)

> Ask several AI models the same question, inspect their individual answers, and use Bearly's synthesis.

**Council of Experts** sends one well-formed question to several AI models in parallel. You can read each answer separately, then use Bearly's synthesis to find consensus, disagreements, and the strongest response.

Council of Experts is available on **Pro** and higher plans. It also requires at least two models that your account and any managed-team policy allow.

## Ask the council



1. **Choose Council of Experts**

   Type `/council` in the message box, then select **Council of Experts** under **AI**. The message box changes to **Ask the council...**.

   If your plan does not include Council of Experts, selecting the command opens the upgrade information instead.

2. **Write a self-contained question**

   Give every expert the same clear objective, important context, and criteria for a good answer. Name an attached file directly instead of relying on “this” or “the document.”

   ```text copy
   Compare the three launch plans in the attached memo. Recommend one for a six-person team with a $40,000 budget, and identify the strongest argument against your recommendation.
   ```

3. **Send the request**

   The **Council of Experts** panel appears in the chat. It shows how many experts are **running**, **done**, **failed**, or **skipped**. Each model has its own tab, with a status dot and its answer as soon as it is ready.

4. **Compare the individual answers**

   Select a model tab to read that expert's response. Use its **Copy** control when you want the answer on its own.

5. **Read the synthesis**

   After enough experts finish, Bearly adds a synthesis beneath the panel. The individual model tabs remain available when you want to compare the source answers with the combined response.



## Choose the models for future councils

After a council panel exists in a chat, select its **Council settings** gear. Under **Council models**, select the lineup you want Bearly to use next time.

- Choose at least 2 and at most 5 experts.
- The selected count appears at the top of the menu.
- Changes save as you make them and affect future councils, not the panel already running or completed.
- The available choices reflect your plan and any model restrictions on a managed account.

If you only want to change one model, deselect it and select its replacement. At the minimum or maximum, Bearly disables choices that would take the council outside the allowed range.

## Understand usage and results

A council uses several model responses and a final synthesis, so it normally consumes more of your Bearly usage than asking one model. The amount varies with the selected models, the number of experts, the length of their answers, and any additional work the question requires.

> **Caution — Treat the council as comparison, not a vote**
> Several models can repeat the same mistake or draw from the same weak premise. For consequential decisions, ask for sources, inspect disagreements, and verify important facts independently.


## If the council cannot finish

| What you see | What to do |
| --- | --- |
| **Council of Experts** does not appear after typing `/council` | Check for a Bearly update. On a managed account, ask an administrator whether Council of Experts is allowed. |
| Selecting the command opens billing | Council of Experts requires Pro or higher. Use the upgrade shown in Bearly to continue. |
| **Council needs more available experts** | Open **Council settings** on an earlier result and select at least two available models. If too few models appear, a team administrator may need to review model access. |
| One expert shows **Error** or **Aborted** | Read the other tabs and the synthesis if at least two experts completed. For a full comparison, start a new council with a different lineup. |
| **Most experts failed** | Check the connection, simplify the question, or choose other models under **Council settings**, then run `/council` again. |
| You stopped the response | Any completed tabs remain readable, but the council may not produce a synthesis. Start a new council when you want to retry. |
| The synthesis misses an important disagreement | Ask a follow-up such as “Compare the experts' assumptions and explain where their recommendations diverge.” |

---

---
title: "Send email or text messages from a chat"
description: "Send formatted email or a concise text message from a Bearly chat, then confirm the result and recover from failures."
canonical: "https://bearly.ai/docs/features/send-email/"
markdown: "https://bearly.ai/docs/features/send-email.md"
---

# Send email or text messages from a chat

HTML: [https://bearly.ai/docs/features/send-email/](https://bearly.ai/docs/features/send-email/)

> Send formatted email or a concise text message from a Bearly chat, then confirm the result and recover from failures.

Ask Bearly to send work from any chat. **Send Email** creates a formatted message for your account address or eligible teammates. **Send SMS** sends a concise plain-text message to the verified phone on your account.

These actions are available on web, desktop, iOS, and Android. An organization policy can hide email or text messaging.

You can also make an explicit request such as “Email this summary to me” in an ordinary message. The slash command makes the sending intent unmistakable. Email is sent when that chat step runs; **Send Email** does not schedule delivery for a later time.

## Send an email



1. **Prepare the material in the chat**

   Ask Bearly to create the summary, draft, checklist, or other content you want to send. Attach or generate any files that should travel with the email before starting the command.

2. **Choose Send Email**

   Type `/email`, then select **Send Email** under **Tools**. The message box changes to **What would you like to email?**

3. **Name the recipients and contents**

   State who should receive the message, the subject, what the body should include, and which chat files to attach. If you do not name a recipient, Bearly sends the email to you.

   ```text copy
   Email me the decision summary from this chat with the subject “Launch review.” Attach the final budget workbook.
   ```

   For a teammate, use the person's name or work email and remove any ambiguity:

   ```text copy
   Email Maya Chen the three approved action items. Use the subject “Monday follow-up,” attach the project brief, and keep the body under 200 words.
   ```

4. **Review the request and send it**

   Check the recipients, subject, requested text, and attachment names before submitting the message. Selecting **Send Email** only stages the command; submitting your request authorizes Bearly to send it without a separate draft-preview screen.

5. **Confirm the result**

   A **Send Email** step appears in the chat. A successful send reports **Email sent to** the recipient or recipient count and shows the subject. Open the email to check the final formatting and attachments.



> **Caution — Sent email cannot be recalled in Bearly**
> After Bearly reports that the email was sent, you cannot edit or unsend it from the chat. Send a corrected follow-up if something needs to change.


## Send a text message

> **Note — Supported phone numbers**
> **Send SMS** is generally available and currently supports verified US and Canadian mobile numbers. An organization policy can turn it off.




1. **Ask for the text**

   In any chat, state what the message should contain. You can type `/sms` and select **Send SMS**, or make an explicit request in an ordinary message.

   ```text copy
   Text me the three launch decisions from this chat. Keep it under 250 characters.
   ```

2. **Set up your phone if Bearly asks**

   If no verified phone can receive the message, the chat shows a **Set up text messages** card. Select **Open Phone & SMS**, enter your number, agree to the texting terms, and enter the code Bearly sends.

   You can also set this up before sending from **Settings** → **Phone & SMS**. After verification, the page shows a masked version of the saved number. You can change, test, opt out, or delete it from the same page.

3. **Send again and confirm the result**

   Return to the chat and ask Bearly to send the text again after setup. A successful **Send SMS** step reports that the text was sent to your verified phone. Open the received message to check the final wording.



Bearly cannot text a number supplied in the chat. It always uses the verified phone on your account. SMS is plain text and does not support chat files or Markdown formatting. Long messages can be rejected because carrier segment limits vary with the characters used.

> **Caution — Sent texts cannot be recalled in Bearly**
> After Bearly reports that the text was sent, you cannot edit or unsend it from the chat. Check whether it arrived before retrying, because a retry can create a duplicate.


## Who can receive an email

- With no recipient specified, Bearly sends to the email address on your account.
- Bearly can search for members whose work email matches your managed organization's domain and send to an exact result. It cannot use this command to send to arbitrary outside addresses.
- A organization administrator can turn the member directory off. When it is off, ordinary members can send only to themselves from chat; organization administrators can still manage the roster.
- When Bearly emails teammates, it also sends a copy to your account address. Replies are directed to that address, so the conversation can continue in ordinary email.
- You can include up to 25 eligible recipients in one email.

If Bearly cannot match a teammate, give it at least two characters from the person's name or work email. Organization administrators can change this access under **Organizations** → your organization → **Overview** → **Settings** → **Member directory**.

## Attach files and stay within limits

You can attach up to 10 files from the current conversation. Each file can be up to 25 MB, with a 50 MB combined limit. Bearly accepts up to 20 email requests from your account in an hour.

The sent confirmation proves that the message was accepted for delivery; it does not list every attachment. A file can be omitted if it is no longer available to Bearly or cannot be prepared for email. Open the received message and verify important attachments.

## If the email does not arrive as expected

| What you see | What to do |
| --- | --- |
| **Send Email** does not appear after typing `/email` | Check for a Bearly update. On a managed account, ask an administrator whether sending email from chat is allowed. |
| Bearly cannot find the teammate | Confirm that the person's membership uses the managed organization's work-email domain. Ask a organization administrator whether the member directory is available, or send the message to yourself and forward it. |
| The **Send Email** step reports a failure | Check the network connection and retry once. If it continues, simplify the message, reduce the attachments, or wait before trying again. |
| The request exceeds an attachment limit | Remove files, make a smaller copy, or split the material across emails, then submit a new request. |
| Bearly says too many emails were sent | Wait until the one-hour limit clears before sending another email. |
| The confirmation appears but the message is missing | Check spam and the inbox for the account or work address shown in your request. Ask Bearly to send a new message only after confirming the address; retries can create duplicates. |
| The email arrived without an attachment | Return to the chat, confirm that the file still opens, attach a smaller or fresh copy if needed, and send a follow-up email naming that file explicitly. |
| The content or recipient was wrong | Send a corrected follow-up. Bearly cannot alter the email that has already been sent. |
| **Send SMS** does not appear | Check for a Bearly update. On a managed account, ask an administrator whether the policy allows text messaging. |
| The chat shows **Set up text messages** | Select **Open Phone & SMS**, complete verification, return to the chat, and submit the request again. |
| Bearly says the text is too long | Ask for a shorter plain-text message without emoji or special characters, then send it again. |
| The text does not arrive | Open **Settings** → **Phone & SMS** and use **Test**. If you opted out, choose **Opt back in** and complete verification before retrying. If the page says texts are off, choose **Re-enable**. |

---

---
title: "Connect services and MCP servers"
description: "Add a supported service or your own MCP server, then choose which chats and Routines can use its tools."
canonical: "https://bearly.ai/docs/features/connectors/"
markdown: "https://bearly.ai/docs/features/connectors.md"
---

# Connect services and MCP servers

HTML: [https://bearly.ai/docs/features/connectors/](https://bearly.ai/docs/features/connectors/)

> Add a supported service or your own MCP server, then choose which chats and Routines can use its tools.

A connector gives Bearly another system's own tools inside a chat — finding a workspace page, reading an issue, checking a project record. You choose what to add, which conversations can use it, and, for a server you add yourself, how Bearly reaches it.

> **Note — Where connectors work**
> HTTPS connectors work in desktop, mobile, and tablet browsers, and in the Bearly desktop app. In the iOS and Android apps, use custom HTTPS servers with a bearer token, custom headers, or no authentication. Services that require an OAuth sign-in need Bearly in a regular web browser or the desktop app. npm package servers require the desktop app.
>
> Connections stay in the profile where you set them up. Signing in to a service in your phone's browser does not connect it in the Bearly mobile app. Managed teams can turn connectors off, prevent members from adding personal MCP servers, or provide MCP servers through a team policy.


## Installed, connected, and active

A connector has three states, and they are not the same thing:

- **Installed** — the connector is in your list in **Connectors**. It is not authorized or checked yet.
- **Connected** — Bearly reached it successfully and a green dot appears beside the name. For a server you added yourself, Bearly also shows how many tools it found.
- **Active** — you turned it on for one conversation.

A connected connector does nothing in a chat until you turn it on there.

## Add a service from the list



1. Open your account menu in the sidebar footer and choose **App Settings**, then select **Connectors**.
2. Under **Available**, find a service and select **Add**. It joins your list but is not authorized yet.
3. Select **Connect**.
4. Complete the service's sign-in and authorization page in the browser. Read the requested permissions before approving them.
5. Return to Bearly. A green dot appears beside the service.



Authorization belongs to the current browser or app profile. On another device, or after signing out of Bearly, connect the service again there. If the service says **Web browser required** in the mobile app, open Bearly in your phone's browser and connect and use the service there.

### Connect Google Workspace

> **Note — Limited availability**
> The Google Workspace connector is available to a limited group while Bearly completes Google's review. It works in interactive chats, not in Routines.


One Google Workspace connection gives Bearly the supported Gmail and Google Calendar actions below.

- **Gmail:** Search email; read messages and threads; list drafts and labels; create drafts; archive a message or thread; and move a message or thread to Gmail Trash. Bearly cannot send or permanently delete email, make other label changes, or download attachment files. It sees attachment names, types, and sizes when Gmail provides them.
- **Google Calendar:** List calendars; find and read events in a time range; and create, update, or delete timed, one-time events without guests. Bearly cannot invite guests, change or delete an event that already has guests, change or delete a recurring event, create an all-day event, create a calendar, or change calendar sharing.

Add and connect Google Workspace using the steps above, then turn it on for a chat. In Standard, when access needs approval, choose **Once**, **This response**, **This session**, or **This chat** in the permission card, then select **Allow** or the button naming your chosen duration. **This response** lets Bearly finish a batch without asking for each item. Session access ends when the chat reloads or the app restarts. Some settings or account-policy changes also reload the chat. Chat access also covers future responses in that chat on the same device until revoked.

[Full access](https://bearly.ai/docs/settings/agent-permissions.md) lets the chat use enabled connector tools without asking each time, including accounts you deliberately connect later. The action-specific confirmations described below apply to Standard.

Creating a Gmail draft has a separate confirmation that shows the recipients and subject. The result remains a draft for you to review and send in Gmail. Read access, draft creation, Archive, and Trash each have separate approvals for the connected account. A longer Archive approval covers more messages and threads without also allowing Trash. Archiving removes the content from Inbox without deleting it; you can find it in **All Mail** and move it back to Inbox. A thread action changes every message in that thread. You can restore content from Gmail Trash until Google permanently removes it under Gmail's retention rules.

Creating, updating, and deleting Calendar events each have separate approvals tied to the connected account and calendar. A longer approval covers additional events in that calendar for the same action. Bearly shows the calendar, event details, and requested change before it writes. A deleted event moves to Google Calendar Trash for 30 days; use Google Calendar on a computer to restore it during that period.

Open **Permissions** near the chat input (or **Full access** while active) to see what is allowed. Choose **Standard** to turn off Full access. Select **Revoke** beside a saved approval, or **Revoke all saved approvals**, to stop current work and ask again. In Standard, disconnecting and reconnecting a service requires new tool approvals. If this chat is open in other windows or tabs, revoke saved approvals in each one.

If you previously connected the connector as **Gmail**, select **Disconnect**, then **Connect**, and approve the updated Gmail and Calendar access once.

Relevant Gmail and Calendar content may be sent to the AI model selected for the chat to complete your request. Bearly treats content returned from either service as data, not as instructions for the assistant.

## Add your own MCP server

Open **App Settings** → **Connectors** and select **Add custom MCP server**. Choose **HTTPS**, or choose **npm package** in the desktop app, give the server a **Name**, then fill in the rest.

Your own servers appear in the same list as the built-in services and work the same way in chats and Routines.

If **Add custom MCP server** is missing but **Connectors** is available, your team policy does not allow personal MCP servers. Built-in services and servers provided by your team remain available.

### An HTTPS server

Enter the **Server URL** — the HTTPS address of a Streamable HTTP MCP server — then choose an **Authentication** option:

- **OAuth (automatic)** for a server that supports MCP sign-in. Bearly opens a sign-in window to approve access. Leave **OAuth scopes** blank to use the server's defaults, or enter the scopes the server documents. This option is available in regular web browsers and the desktop app, not the iOS or Android apps.
- **Bearer token** for a server that expects a token. Give the value a local secret name and enter the token.
- **Custom headers** for a server that expects an API key or another named header. Enter the header name, a local secret name, and the value.
- **No authentication** for a public server.

Bearly attaches these headers only to requests to the server address you entered.

### An npm package

Enter the package name under **npm package**, optionally with a version or tag, such as `@scope/mcp-server@latest`. Add **Arguments** one per line if the package needs them, and select **Add variable** for each environment variable it expects.

npm package servers can run only in the Bearly desktop app. Bearly runs the package with a runtime it installs and manages, so you do not need Node or npm on your computer. The first check can take a minute while it prepares. The package then runs locally, under your user account.

### Finish and check

Select **Add and check**. Bearly saves any values you entered, contacts the server, and shows the number of tools it found. If it cannot connect, the reason appears in red under the row — fix it and select **Connect** to try again.

### Where authentication values are kept

Values for an HTTPS server are stored in the current Bearly browser or app profile, separately from desktop process environment variables. They are attached only to requests for the server address you entered. Those requests pass through Bearly's authenticated HTTPS proxy so servers that do not allow browser CORS requests still work.

Values for an npm package are stored in this device's environment variables at **App Settings** → **Runtime**. The same variables are available to other local features on this device, such as skills and local commands.

Change an HTTPS value by editing that connector. Change an npm value in the connector or under **Runtime** → **Environment variables**.

> **Caution — Only add servers you trust**
> An npm package runs on your computer as you and can reach files your account can reach. A remote server receives whatever a tool call includes. Add a server because you know its publisher, not because its name sounds useful.


## Use an MCP server provided by your team

A team admin can add HTTPS or npm MCP servers to a policy. When that policy applies to you, compatible servers appear automatically in **App Settings** → **Connectors** with **Managed by your team policy** under their names. npm servers appear only in the desktop app. You cannot change a managed server's address, package, arguments, authentication method, or required variable names.

Authentication is still yours:

- For OAuth, select **Connect** and approve access with your own account in a regular browser or the desktop app.
- For a bearer token or custom header, select the pencil beside the server and enter your value. The value stays in this browser or app profile.
- For an npm server, open the Bearly desktop app and enter the required environment values. The package cannot run in a browser.

If your assigned policy changes or an admin removes the server, Bearly removes that managed server from your available Connectors. It does not remove a server you added personally.

## Turn a connector on for one chat



1. Open the conversation and select **Connectors** in the context strip beside the message box.
2. Select a connected service. Its row changes from **Off** to **Active**, and the chip shows the service name or the number of active connectors.
3. Ask Bearly for a specific result, including any boundary that matters.

   ```text copy
   Using the connected project service, find open issues assigned to me that changed this week. Summarize them; do not edit anything.
   ```

4. Open **Connectors** again and select the service to turn it **Off** for that chat.



A service marked **Connect** in this menu can be connected from here; Bearly makes it active in that conversation once the connection succeeds. A row marked **Retry** failed the last time it was checked. **Manage connectors...** at the bottom opens settings.

In Standard, when Bearly asks to use a connected server, choose **Once**, **This response**, **This session**, or **This chat**. A longer approval covers that server's tools, including actions that change data. You can revoke it under **Saved approvals** in the chat's **Permissions** control.

The exact abilities come from the connected service and your account permissions there. If your account cannot see or change something, Bearly cannot bypass that restriction.

## Use a connector in a Routine

In the routine editor, select **Plugin**, then pick the connector under **Connectors**. Its chip has a **Tools** setting: **All**, or **Read-only** to keep the run to tools that only look things up.

Routines run on the device where you set them up, using that device's connections. If a selected connector is not connected there, the run says so rather than failing quietly. See [Run a prompt on a schedule](https://bearly.ai/docs/features/routines.md).

Google Workspace does not appear in the Routine connector list. Approvals granted in an interactive chat do not authorize unattended Routine access.

## Edit, disconnect, or remove

Open **App Settings** → **Connectors**:

- Select the pencil **Edit** icon beside a server you added to change its settings. Saving re-checks the server.
- For a server marked **Managed by your team policy**, select the pencil to add or replace your local authentication values. Its configuration and removal are controlled by your team admin.
- Select **Disconnect** to drop the current authorization or connection while keeping the connector installed. Select **Connect** when you want it back.
- Select the **Remove** X to take it out of your list. For a local package server, this also stops it.

Disconnecting or removing a connector in Bearly does not delete anything in the outside service. To also revoke Bearly's access at the source, use that service's connected-app or security settings. For Google Workspace, remove Bearly from your [Google Account connections](https://myaccount.google.com/connections).

## Privacy and trust

Anything a connector returns becomes part of the task and may be sent to the AI model producing the answer. Apply the same care you would use when pasting material from that service into a chat.

Servers you add personally stay in the browser or app profile where you added them. Their address, package, variable names, and values are not copied to your other devices or profiles, so set the server up again anywhere else you want it.

For a server provided by your team, the policy copies the server definition and required secret or environment variable names to members. It never includes the corresponding token, header, OAuth authorization, or environment value. Enter those values separately on each browser or app profile where you use the server.

> **Caution — State the boundary for consequential work**
> An active connector may offer both reading and writing actions. For sensitive records, payments, deployments, publishing, or deletion, begin with an inspection-only request and review the result before asking Bearly to make a change.


## If a connector does not work

| What you see | What to do |
| --- | --- |
| **Connectors** is missing | Check whether your managed-team policy permits connectors and whether you are using a current Bearly version. |
| A service says **Web browser required** | Open Bearly in a regular web browser and connect and use the service there. Its sign-in cannot complete inside the mobile app, and the connection is not transferred between profiles. |
| An npm server is missing on a phone or in a browser | Open the Bearly desktop app. npm servers run locally and are available only there. |
| **Add custom MCP server** is missing | Your team policy allows Connectors but not personal MCP servers. Use a built-in or team-provided server, or ask your team admin to change the policy. |
| A server says **Managed by your team policy** | Select the pencil to enter your own authentication values. Ask your team admin to change its address, package, arguments, or required variable names. |
| Authorization closes or is cancelled | Return to **App Settings** → **Connectors** and select **Connect** again. Complete the entire browser flow with the intended account. |
| A red dot or connection error appears | Read the message under the row, select **Connect** to retry, and reauthorize if the earlier permission expired or was revoked. |
| `Set NAME in Desktop → Runtime.` | The connector refers to an environment variable that no longer exists. Add it under **App Settings** → **Runtime**, or edit the connector to point at a variable you have. |
| `Edit this connector and enter a value for NAME.` | An HTTPS server is missing a token or header value. Select the pencil beside it and enter the value again. |
| A server you added will not connect | Confirm the address is a public HTTPS MCP endpoint, or that the package name exists in the npm registry. Check that the token or key is current, then select **Connect** again. |
| A service is connected but Bearly does not use it | Open the chat's **Connectors** menu and confirm the service says **Active**, then name the service in the request. |
| Bearly cannot find a record you can see | Confirm you authorized the correct account or workspace and that it has access to the record. |
| A tool fails partway through a task | Ask Bearly what completed, verify the outside service directly, and retry only the failed step. Do not assume a write was rolled back. |
| A routine reports a connector is not connected | Open **Connectors** on the device that runs the routine and connect it there. |

---

---
title: "Organize chats with projects"
description: "Keep related chats, files, instructions, and settings together in a Bearly project."
canonical: "https://bearly.ai/docs/projects/overview/"
markdown: "https://bearly.ai/docs/projects/overview.md"
---

# Organize chats with projects

HTML: [https://bearly.ai/docs/projects/overview/](https://bearly.ai/docs/projects/overview/)

> Keep related chats, files, instructions, and settings together in a Bearly project.

Projects give a body of work its own place in Bearly. A project can hold related chats and files, apply the same chat defaults across those conversations, and be shared with people on your teams.

## Create a project



1. In the chat sidebar, find **Projects** and select the **New project** folder-plus button.
2. In **Create New Project**, enter a **Project Name**.
3. Optional: turn on **Enable password encryption?**, then enter and confirm a password of at least eight characters.
4. Select **Create Project**. Bearly opens the new project's **Settings** page.



Your organization may require password protection or prevent project creation. In that case, the password option is already on or the **New project** button is unavailable.

You can also choose a project while starting a chat. Select the **No project** chip below the new-chat composer, type in **Find or create a project...**, then choose an existing project or **Create “project name”**. The chat starts in that project when you send its first message.

## Work inside a project

Select a project name in the sidebar to expand it. Its main destinations are:

- **Settings** — set project-wide chat defaults, manage access, add Skills, protect the project, and—on desktop—sync a local directory.
- **Docs & Data** — upload, search, inspect, and remove files available to project chats.
- **Start chat** — begin a new conversation in the project.

Project chats appear beneath those controls. Select the project-name chip in an open chat to open that project's **Docs & Data**.

To control conversational memory within a project chat, select the sliders icon beside the message box and open **Memories**. Choose **Project Only** to recall context from chats in the same project, **All Chats** to search more broadly, or **Off** to avoid recalling earlier conversations.

### Move a chat into or out of a project

Open the chat's three-dot menu and select **Move to project...**, then choose a project. The picker puts pinned and recently used projects first and includes **Search projects...** when you need it.

To return a project chat to the main chat list, open its three-dot menu, select **Move to project...**, then select **Remove from project**. Removing a chat from a project does not delete the chat.

On a computer, you can also drag a chat onto a project or drag it back to the main chat list. Use **Move to project...** on touch devices.

## Rename, archive, or delete a project

Open the project's three-dot menu for these actions. **Rename** and **Delete** are available to project owners.

### Rename

Select **Rename**, enter the new name, then press <kbd>Enter</kbd> or click away. Bearly currently uses the standard folder icon for projects; the project menu does not offer custom colors or icons.

### Archive

Select **Archive** to hide a project from the regular sidebar without deleting it. When you have archived items, select **Archives** in the sidebar footer to open the list of everything archived, grouped by archive date. Select the project there to see its chats, or open its menu and select **Unarchive** to restore it.

An archived project cannot be deleted until you unarchive it.

### Delete



1. Open the project's three-dot menu and select **Delete**.
2. Decide whether to turn on **Also delete all chats in this project**.
   - Off: the project's chats return to the main chat list.
   - On: chats in the project are permanently deleted with it.
3. Select **Delete Project**.



> **Danger**
> Deleting a project and its settings cannot be undone. Archived chats are protected from deletion; if you need to delete them too, unarchive them first.


## Continue setting up the project

- [Add files to a project](https://bearly.ai/docs/projects/documents.md)
- [Set project-wide chat defaults](https://bearly.ai/docs/projects/settings.md)
- [Share a project with your team](https://bearly.ai/docs/projects/collaboration.md)
- [Protect a project with a password](https://bearly.ai/docs/projects/encryption.md)

---

---
title: "Add files to a project"
description: "Upload, search, inspect, and manage files that Bearly can use across a project's chats."
canonical: "https://bearly.ai/docs/projects/documents/"
markdown: "https://bearly.ai/docs/projects/documents.md"
---

# Add files to a project

HTML: [https://bearly.ai/docs/projects/documents/](https://bearly.ai/docs/projects/documents/)

> Upload, search, inspect, and manage files that Bearly can use across a project's chats.

Project files give every chat in a project access to the same source material. They are useful for a client brief, a research library, a set of reports, or any collection you expect to revisit across conversations.

## Upload files

Writers and owners can add and remove project files. Readers can view the files and use them in project chats. Your organization may turn project files or uploads off.



1. Expand the project in the chat sidebar and select **Docs & Data**.
2. Select **Upload** and choose one or more files. You can also drag files onto the page on a computer.
3. Keep the page open while the **Uploading** area shows each file as **Waiting...**, a percentage, **Complete**, or **Failed**.
4. After upload, Bearly analyzes supported content so it can be found and used in project chats. A spinner beside a file means analysis is still running.



If the project is empty, **Upload Your First File** opens the same file picker. You can cancel an upload in progress or dismiss a failed upload from its row.

### What works best

Document search is best for PDFs, Word documents, presentations, Markdown, plain text, and similar text-rich files. CSV and spreadsheet content can be searched as extracted text, but Bearly does not preserve it as an interactive table during document search.

Images, audio, video, and other files can still live in the project and may have their own preview or analysis tools, but they are not searched as ordinary document text. Bearly keeps the original file even when it cannot fully preview or analyze its contents.

## Find and inspect a file

Use **Search files...** to filter by filename. Select **Grid view** or **List view** to change the layout.

Select a file to open **File Details**. Depending on the file and completed analysis, this panel can show:

- its name, size, type, and date added;
- a description, topics, people, date, or amount of extracted text;
- the status of each analysis step; and
- directory-sync information for files added from a desktop folder.

A warning icon means analysis failed or is incomplete. Open **File Details** to read the user-facing error. If **Rerun Analysis** appears, select it to try again. **Get help** opens a support request with the relevant diagnostic details.

> **Note**
> When semantic search is unavailable for a file, Bearly may still be able to inspect it directly. For reliable answers across a large set, resolve the warning or use a more text-friendly version of the file.


## Use project files in a chat

Start or open a chat inside the project and ask a specific question, naming a file when that helps. For example:

> Compare the renewal terms in the 2025 and 2026 agreements, and cite the relevant passages.

Project files are available to the chat without attaching them again. To make every response consult the collection, expand the project, select **Settings**, enable **Project Chat Settings**, and turn on **Always Search Documents**. Leave it off when most questions do not depend on the files.

Files that you explicitly ask Bearly to save to the project also appear in **Docs & Data**.

## Delete files

Deleting a project file is permanent.

- For one file, open **File Details**, select **Delete File**, then confirm **Delete**.
- For several files, select **Enter Select Mode**, choose the files or **Select All**, then select **Delete (number)** and confirm **Delete**.

If the delete controls do not appear, you have read-only access or your organization's policy prevents changes to project files.

## Sync a desktop folder

The Bearly desktop app can keep a local directory in sync with a project. Open the project's **Settings** page and find **Local Directory Sync**, then select **Add Directory**.

Choose **Browse**, review **Auto-remove**, **Max file size**, and the optional **File types** filter, then select **Create Sync**. Bearly shows the source as **Pending**, **Syncing**, **Synced**, or **Error** and identifies which device owns it.

**Auto-remove** removes the project copy when the source file disappears from the directory. Deleting the sync itself leaves files that were already added in the project. Directory sync requires the desktop app, writer access, and permission to add project files.

Related: [Set project-wide chat defaults](https://bearly.ai/docs/projects/settings.md)

---

---
title: "Set project-wide chat defaults"
description: "Choose the model, language, style, instructions, and document behavior for chats in a project."
canonical: "https://bearly.ai/docs/projects/settings/"
markdown: "https://bearly.ai/docs/projects/settings.md"
---

# Set project-wide chat defaults

HTML: [https://bearly.ai/docs/projects/settings/](https://bearly.ai/docs/projects/settings/)

> Choose the model, language, style, instructions, and document behavior for chats in a project.

Project chat settings keep related conversations consistent. They affect future messages in a project's chats; they do not rewrite messages that are already in the conversation.

## Open project settings

Expand the project in the chat sidebar and select **Settings**. Find **Project Chat Settings**, then turn it on. In a shared project, this section is named **Shared Project Settings**.

Changes save automatically. The section's **Enabled** or **Disabled** badge tells you whether its defaults are active.

## Choose the defaults

Set only the fields that should be consistent across the project:

- **Model** — chooses the default from the models available to your account.
- **Language** — sets the language Bearly should use in its replies.
- **Response Style** — applies a response preset, or **Default** for the standard style.
- **Temperature** — adjusts how focused or varied responses can be. Use **Factual** (0.5), **Default** (1.0), **Creative** (1.5), or the slider from 0 to 2. Values above 1.5 can produce incoherent output.
- **Instructions** — gives Bearly durable guidance for this project. Include the goal, audience, important constraints, and preferred output—not secrets or information better kept in a source file.
- **Disable Personalization** — ignores your account-level personalization in this project.
- **Always Search Documents** — searches project documents before responding. This is useful when most questions depend on the project's source collection, but it can add unnecessary work to unrelated questions.

For example, project instructions might say:

> Write for an informed general reader. Distinguish facts from assumptions, preserve cited figures exactly, and flag conflicts between the source documents.

## Control individual overrides

The lock beside **Model**, **Language**, **Response Style**, **Temperature**, **Instructions**, and **Always Search Documents** controls whether those settings can be changed later for an individual chat.

- An open-lock icon means a chat can override the project default.
- A closed-lock icon means users cannot change that field in a chat.

Lock only the settings that truly need to remain consistent. A team policy can also disable or lock a field; when that happens, the setting is dimmed and marked **Disabled by policy.**

## Settings in a shared project

Writers and owners can change **Shared Project Settings**. Readers can use the project but cannot change its shared defaults.

Each collaborator also gets **Personal Chat Settings**. When enabled, these preferences apply only to that person and can override unlocked shared defaults. A shared lock or team policy still takes precedence.

The **Skills** section controls Skills available to chats in the project. Writers and owners can change the list when their account policy permits it; other collaborators can still use shared Skills that are already attached.

## Sync a local directory on desktop

In the Bearly desktop app, writers and owners may also see **Local Directory Sync** on this page. Select **Add Directory**, then:

1. Select **Browse** and choose a folder.
2. Decide whether **Auto-remove** should remove project copies when their source files disappear.
3. Set **Max file size** and, if useful, a comma-separated **File types** filter such as `.md, .txt, .pdf`.
4. Select **Create Sync**.

Use **Sync now** to run a source immediately or **Configure sync** to change its filters. Removing a sync does not remove files already copied into the project. This section is unavailable outside the desktop app or when project-file changes are restricted.

## If a setting does not take effect

- Confirm the section says **Enabled**.
- Check whether the field is marked **Disabled by policy.**
- In a shared project, review **Personal Chat Settings** for an unlocked personal override.
- Confirm the chat is actually inside the project—the project-name chip appears below its composer.
- For document-grounded answers, confirm the files have finished processing in **Docs & Data**.

Related: [Add files to a project](https://bearly.ai/docs/projects/documents.md)

---

---
title: "Share a project with your team"
description: "Give a team or individual teammates Reader, Writer, or Owner access to a Bearly project."
canonical: "https://bearly.ai/docs/projects/collaboration/"
markdown: "https://bearly.ai/docs/projects/collaboration.md"
---

# Share a project with your team

HTML: [https://bearly.ai/docs/projects/collaboration/](https://bearly.ai/docs/projects/collaboration/)

> Give a team or individual teammates Reader, Writer, or Owner access to a Bearly project.

Project collaboration lets a group work from the same chats, files, and shared project settings. You can share only with teams you belong to and people listed in those teams.

## Share the project

You must be a project owner and belong to at least one team. Your organization may disable project collaboration.



1. Expand the project in the chat sidebar and select **Settings**.
2. In **Collaboration**, select **Share**.
3. To share with everyone on a team, choose a role from the selector on that team's row.
4. To share with particular people, select the **Show members** chevron for the team, then choose a role beside each person.
5. Select **Done** when the access list is correct.



The project status changes from **Private** to **Shared**, and a people icon appears beside the project in the sidebar. Return to the same section and select **Manage** whenever you need to review access.

If no teams are available, create or join a team before sharing the project.

## Choose an access level

The role selector offers four levels:

- **No Access** — removes access granted directly through that team or person.
- **Reader** — can view the project and participate in its existing chats.
- **Writer** — has Reader access and can start chats, add or remove project files, change shared chat settings, and manage project Skills when allowed.
- **Owner** — has full project control, including sharing, password protection, renaming, and deletion.

> **Caution**
> Give **Owner** access sparingly. An owner can change other people's access and permanently delete the project.


A person can receive access from both a team and an individual assignment. Selecting **No Access** on one row removes that assignment; it does not cancel access they still receive through another team or owner role.

## Work in a shared project

Shared project behavior depends on the role:

- **Shared Project Settings** contain defaults for everyone. Writers and owners can change them; Readers cannot.
- **Personal Chat Settings** let each collaborator set their own unlocked preferences without changing the shared defaults.
- Project files in **Docs & Data** are available to all collaborators. Writers and owners can add and remove them unless an organization policy prevents it.
- A password-protected project still requires its project password after it is locked. Share that password through an appropriate secure channel, separate from Bearly.

If a role update fails, Bearly leaves an error in the collaboration panel. Try again; if it continues, confirm you are still a project owner and that the teammate still belongs to one of your teams.

Related: [Set project-wide chat defaults](https://bearly.ai/docs/projects/settings.md) and [Protect a project with a password](https://bearly.ai/docs/projects/encryption.md)

---

---
title: "Protect a project with a password"
description: "Add, use, change, or remove password protection for a Bearly project."
canonical: "https://bearly.ai/docs/projects/encryption/"
markdown: "https://bearly.ai/docs/projects/encryption.md"
---

# Protect a project with a password

HTML: [https://bearly.ai/docs/projects/encryption/](https://bearly.ai/docs/projects/encryption/)

> Add, use, change, or remove password protection for a Bearly project.

Password protection adds a separate lock to a project's files and settings, and makes its chats read-only while the project is locked. Only project owners can manage it.

> **Danger**
> Bearly cannot recover a protected project if you forget its password. Store the password somewhere secure before you continue.


## Protect an existing project



1. Expand the project in the chat sidebar and select **Settings**.
2. In **Encryption**, select **Protect**.
3. Enter a **Password (min 8 characters)**, then enter it again in **Confirm password**.
4. Select **Protect**. The status changes from **Not Protected** to **Protected**.



You can also protect a project while creating it: turn on **Enable password encryption?** in **Create New Project**, then enter and confirm a password of at least eight characters.

Some managed accounts require a password for every project. In that case, protection is already enabled in the creation form and cannot be removed later.

## Lock and unlock the project

While a protected project is unlocked, open its **Settings** page and select **Lock** under **Encryption**. Project chats become read-only, and opening **Settings** or **Docs & Data** shows the password screen.

Open **Settings** or **Docs & Data**, enter the password, and select **Unlock**. If you reuse the same password for several protected rooms, turn on **Unlock all rooms that use this password?** before unlocking to try that password on the others as well.

If Bearly reports **Incorrect password. Please try again.**, re-enter the password carefully. There is no password-reset or recovery path.

## Change the password



1. Open the unlocked project's **Settings** page.
2. Under **Encryption**, select **Change**.
3. Enter the **Current password**.
4. Enter a **New password (min 8 characters)** and **Confirm new password**.
5. Select **Change Password**.



Anyone else who uses the protected project will need the new password after the project is locked again.

## Remove password protection



1. Open the unlocked project's **Settings** page.
2. Under **Encryption**, select **Remove**.
3. Read the warning, then select **Remove Protection**.



Removing protection is permanent until you add a new password. People who already have project access—and anyone using your signed-in account—will no longer be asked for the project password.

If **Remove** is unavailable, your organization's policy may require project passwords. If the status reads **Device Encrypted**, the project uses a device-based account policy rather than a password you can change here; contact your account administrator or Bearly support if you cannot open it.

Related: [Share a project with your team](https://bearly.ai/docs/projects/collaboration.md)

---

---
title: "Create and manage an organization"
description: "Create a Bearly organization, review its account details, rename it, leave it, or delete it when eligible."
canonical: "https://bearly.ai/docs/teams/overview/"
markdown: "https://bearly.ai/docs/teams/overview.md"
---

# Create and manage an organization

HTML: [https://bearly.ai/docs/teams/overview/](https://bearly.ai/docs/teams/overview/)

> Create a Bearly organization, review its account details, rename it, leave it, or delete it when eligible.

Organizations give a group of people a shared place to collaborate on chats and projects. What you can change depends on your role: an **Admin** manages the organization, while a **Member** can review the organization and their own membership.

## Create an organization



1. Open your account menu and select **Organizations**.

2. Start the organization:

   - If you see **No organizations yet**, enter a name in **Organization name**.
   - If you already belong to an organization, select **New organization**, then enter a name in the **New organization** dialog.

   Organization names must contain 4–18 characters.

3. Select **Create** for your first organization, or **Create organization** in the dialog.



Bearly opens the new organization's **Overview** and marks you as an **Admin**. If **New organization** does not appear, your account cannot create another organization; ask an organization admin for help.

## Open and understand an organization

Open your account menu, select **Organizations**, then select the organization name. The badge beside the page title shows whether you are an **Admin** or **Member**.

Every organization includes **Overview**, **Members**, and [**Groups**](https://bearly.ai/docs/teams/groups.md). Administrators also see **Policies** and **Activity**. **Billing** appears when the organization can buy seats directly and does not already receive a plan through another arrangement. Additional sections may appear for some managed organizations.

The **Overview** shows the organization's name and plan, plus its managed email domain when it has one. Administrators also see the member count, managed organization policy count, a shortcut to seat billing when it is available, and the organization's **Settings**. They can also [set an access banner](https://bearly.ai/docs/teams/access-banner.md) that members must accept before using Bearly. On a managed organization, a **Microsoft Entra** section follows **Settings** for [Microsoft sign-in and provisioning](https://bearly.ai/docs/teams/microsoft-entra.md). Members instead see **Your membership**, including their role and assigned policy when one applies.

To move between organizations, select **Organizations** and choose another organization name.

## Rename an organization

Only organization administrators can rename an organization.

1. Open the organization and select **Overview**.
2. Select the **Rename organization** pencil beside **Organization name**.
3. Enter a name containing 4–18 characters.
4. Select the check mark or press <kbd>Enter</kbd> to save. Select the X or press <kbd>Esc</kbd> to cancel.

If Bearly cannot save the change, the existing name remains in place. Try again before making other organization changes.

<span id="request-a-managed-team"></span>

## Request a managed organization

An administrator of an eligible free organization may see **Upgrade to Managed** beside the plan.

1. Select **Upgrade to Managed**.
2. Choose **Pro**, **Analyst**, or **Research** under **Plan**.
3. Optionally enter your organization's email domain under **Domain**. If the request is approved with a managed domain, people using that domain are automatically added to the organization.
4. Select **Send Request**.

After the request is sent, **Request is being reviewed** appears beside the plan. This is a request for Bearly to configure the organization; it does not change the plan immediately.

## Leave an organization

Members of an unmanaged organization can leave it themselves:

1. Open the organization and select **Overview**.
2. Under **Your membership**, select **Leave organization**.
3. Confirm **Leave organization**.

Leaving removes your organization membership and your access to its shared chats and projects. Members whose account is managed by the organization's email domain cannot leave themselves; **Overview** identifies the domain managing their membership.

The last organization administrator cannot leave or remove their own membership. Make another member an organization administrator first.

## Delete an organization

Only an administrator can delete a free, unmanaged organization. Managed organizations and organizations with a paid plan must be handled through Bearly Support.

> **Danger**
> Deleting an organization permanently deletes the organization and all of its associated data. This cannot be undone.




1. Open the organization's **Overview** and scroll to **Settings**.

2. Expand **Delete organization**.

3. Enter the exact organization name in both **Confirm name** and **Confirm name again**.

4. Select **Delete organization**.



Bearly returns you to **Organizations** after deletion. If the organization is managed or has a paid plan, the deletion form is replaced by instructions to contact support.

## Next step

Learn how to [add, edit, and remove organization members](https://bearly.ai/docs/teams/members.md) or [buy and manage organization seats](https://bearly.ai/docs/teams/billing.md).

---

---
title: "Set an organization access banner"
description: "Require organization members to accept a custom message before using Bearly."
canonical: "https://bearly.ai/docs/teams/access-banner/"
markdown: "https://bearly.ai/docs/teams/access-banner.md"
---

# Set an organization access banner

HTML: [https://bearly.ai/docs/teams/access-banner/](https://bearly.ai/docs/teams/access-banner/)

> Require organization members to accept a custom message before using Bearly.

An access banner shows organization members a message they must accept before they can use Bearly. Only an organization **Admin** can configure it.

## Configure the banner



1. Open your account menu and follow **Organizations** → your organization → **Overview**.

2. Find **Access banner** and select **Configure**.

3. Turn on **Show access banner**.

4. Enter the **Heading** and **Message (Markdown)** that members should see.

5. Set the **Accept button text**. The default is **Accept & Continue**.

6. To give members a way to leave without accepting, turn on **Let members sign out**. This adds a **Sign out** button; its text and action cannot be changed.

7. Choose a **Frequency**:

   - **Session** — once each time the app or browser is opened.
   - **Daily** — once every 24 hours after acceptance.
   - **Weekly** — once every 7 days after acceptance.

8. Review the preview as you edit. Select **Open preview** to see the full banner.

9. Select **Save banner**.



Bearly confirms with **Access banner saved**. The collapsed setting shows **Enabled** and the selected frequency.

> **Note**
> Acceptance is remembered separately in each browser or app installation. A member may be asked again after switching devices, using another browser, or clearing local app or browser data.


## What members see

The banner appears before Bearly can be used. A member must select the accept button to continue. If **Let members sign out** is on, they can select **Sign out** instead.

Saving changes to an enabled banner creates a new version. Members are asked to accept the updated banner even if their previous acceptance would otherwise still be current.

## Edit or turn off the banner

Return to the organization's **Overview**, find **Access banner**, and select **Configure**.

- To change the banner, edit its settings and select **Save banner**.
- To discard unsaved edits, select **Discard changes**.
- To stop showing the banner, turn off **Show access banner** and select **Save banner**. The collapsed setting then shows **Off**.

If **Save banner** is unavailable, add a heading and accept-button label, and add a message before turning the banner on. The heading can contain up to 120 characters, the accept-button label up to 80 characters, and the Markdown message up to 20,000 characters.

Related: [Create and manage an organization](https://bearly.ai/docs/teams/overview.md) and [Add and manage organization members](https://bearly.ai/docs/teams/members.md)

---

---
title: "Manage organization members"
description: "Add people to a Bearly organization, import a roster from CSV, change member access, and remove members."
canonical: "https://bearly.ai/docs/teams/members/"
markdown: "https://bearly.ai/docs/teams/members.md"
---

# Manage organization members

HTML: [https://bearly.ai/docs/teams/members/](https://bearly.ai/docs/teams/members/)

> Add people to a Bearly organization, import a roster from CSV, change member access, and remove members.

Ordinary members can view their own membership. Organization administrators can view the complete roster, add people, change roles and managed organization settings, and remove members. A managed organization can also receive members and groups automatically from [Microsoft Entra](https://bearly.ai/docs/teams/microsoft-entra.md).

Open your account menu and follow **Organizations** → your organization → **Members**. Use **Search members…** to find someone by name or email.

## Add one member



1. Select **Add member**.

2. Enter the person's **Name** and **Email**.

3. Leave **Organization admin** off for an ordinary member. Turn it on only when the person should have full management permissions across the organization.

4. For someone whose email matches a managed organization's domain, choose a **Policy** when policies are available. The organization's default policy is selected initially.

   You can also enter a monthly **Limit**. This member-specific limit overrides the policy limit; leave it blank when no individual limit is needed.

5. Select **Add Member**.



The person becomes an active organization member immediately and appears in the member list. Bearly does not send or wait for an organization invitation to be accepted.

> **Note**
> A managed organization only accepts members whose email matches its managed domain. Policy and spending-limit controls appear only for those managed accounts.


## Add members from a CSV

Use a CSV import when you already have a roster.



1. Select **Bulk add members**, then **Select CSV File**.

2. Upload a `.csv` file with exactly two columns: name and email. Either order works:

   ```text
   Name, Email
   Ada Lovelace, ada@example.com
   ```

   You may include a header row; Bearly detects it automatically.

3. Review **Ready to Add Members**. Bearly separates the file into **To add** and **Already exist**, and shows the current member count.

4. Leave **Sync Mode** off when you only want to add people.

5. Select **Add _N_ Members** and wait for **Complete**. Review the **Added**, **Skipped**, and **Errors** counts before selecting **Done**.



CSV imports add people as ordinary members. On a managed organization, they receive the default policy. You can promote an administrator, choose a different policy, or set an individual limit afterward.

> **Caution**
> **Sync Mode** treats the CSV as the organization roster. Members who are not in the file are shown for removal and will be removed if you continue; Bearly does not include your own account in that removal list. Review the list carefully before selecting **Add _N_ & Remove _N_** or **Remove _N_ Members**.


If Bearly rejects the file, correct every reported row and upload it again. Common problems include extra columns, missing names or email addresses, and invalid email formats. If some operations fail after processing, the results list identifies the affected people; a second import will skip anyone already added.

## Control teammate contact lookup

On a managed organization, open **Organizations** → your organization → **Overview**, then find **Member directory** under **Settings**. Choose one of these options:

- **All managed members** lets ordinary members ask Bearly to find a managed-domain teammate by name or work email for features such as **Send Email**.
- **Disabled for members** limits ordinary members to sending email to themselves from chat.

This setting exposes only the names and work email addresses needed for contact lookup. It does not change the full roster available to organization administrators or reveal member roles, policies, spending limits, or other membership settings.

## Edit a member

Only organization administrators can change member settings.

1. Select the member's row or their **Member settings** gear.
2. Change their **Name** if needed. The **Email** is read-only; remove and re-add the member to correct an email address.
3. Use **Organization admin** to grant or remove full organization-management access.
4. For a managed account, choose a **Policy** and, if needed, enter a **Monthly spend limit**. Leave the limit blank to remove the member-specific limit.
5. Select **Save**.

Bearly will not let you remove administrator access from the organization's last administrator. Make another member an **Organization admin** first.

If a policy badge includes **[Inactive]**, that member still references a policy that is no longer active on the organization. Open their settings, choose an active policy or **No Policy**, then select **Save**.

## Remove a member

> **Caution**
> Removing a member ends their organization membership and access to the organization's shared chats and projects. If they hold an organization seat, Bearly unassigns it and leaves the paid seat open for someone else. The removal cannot be undone, although an administrator can add the person again later.


1. Open the member's row with **Member settings**.
2. Select **Remove**.
3. Review the confirmation, then select **Remove Member**.

The last organization administrator cannot be removed. Make another member an **Organization admin**, then remove the original administrator.

If the wrong person was removed, use **Add member** to add them again and restore the appropriate role, policy, and individual limit.

---

---
title: "Manage groups and group admins"
description: "Organize departments within a Bearly organization, delegate group management, set policies, and review shared projects and usage."
canonical: "https://bearly.ai/docs/teams/groups/"
markdown: "https://bearly.ai/docs/teams/groups.md"
---

# Manage groups and group admins

HTML: [https://bearly.ai/docs/teams/groups/](https://bearly.ai/docs/teams/groups/)

> Organize departments within a Bearly organization, delegate group management, set policies, and review shared projects and usage.

Use groups for departments and teams within one Bearly organization. A group can have its own members, assigned policy, administrators, and shared projects. Prepaid organizations can use these controls without buying or assigning seats.

## Open a group

Open your account menu → **Organizations** → organization name → **Groups**, then select a group.

Organization admins can manage every group. A **Group admin** manages only their assigned groups. Other organization members see **Shared projects**, limited to projects they can already access.

To create a group, an organization admin selects **Add group**, enters a **Group name**, and selects **Add group** again. The group appears in the list; add members next.

A **Synced** group comes from your identity provider. Its name and synced membership are managed there. A synced group and a separate Bearly organization with the same name remain separate; matching names do not move members or shared projects between them. See [Microsoft Entra provisioning](https://bearly.ai/docs/teams/microsoft-entra.md) for setup.

## Add members and appoint admins

People must already belong to the organization before you add them to a group.

1. Select the group and open **Members**.
2. As an organization admin, use **Find an organization member to add…**, then select **Add** beside a result. As a group admin, enter the person's exact **Organization member email** and select **Add member**.
3. Confirm that the person appears in the group's member list.

Only an organization admin can select **Make group admin** beside a member. The **Group admin** badge confirms the role. Select **Remove admin role** to revoke it. Granting this role does not make the person an organization admin or give them additional access to project content.

Select **Remove** to remove a manual membership. Group admins can remove ordinary manual members; organization admins handle group-admin membership removal. For a synced-only member, **Remove** is unavailable: change the membership in the identity provider. If both manual and synced membership exist, removing the manual membership preserves the synced membership.

## Set policy and group settings

Open the group's **Policy and settings** section.

Organization admins can choose a **Policy**, set its **Priority**, switch **Active** on or off, and select **Save**. A direct member policy takes precedence. Otherwise, the highest-priority active group policy applies. Priorities for active groups with assigned policies must be distinct.

Group admins can rename manually created groups and customize an assigned policy. An organization admin must assign the initial policy first.

1. Select **Customize group policy**.
2. Adjust the policy settings.
3. Select **Save** in the policy editor.

Customization creates a policy specific to this group. It preserves the shared policy used by other groups; later changes to that shared policy do not update this customized copy. Organization admins retain control of **Monthly spend limit**, which applies per managed-domain member. Group admins can see it but cannot change or clear it. A member-specific spending limit still takes precedence.

An organization admin can deactivate a group or delete a manually created group. Deactivation ends access and policy application through that group. Deletion also removes the group itself. Synced groups are removed through the identity provider. Other organization or individual project permissions still apply.

## Review usage

For a managed organization, organization admins and the group's admins can open the group's **Usage** section. Choose **1 mo**, **3 mo**, **6 mo**, or **12 mo**, then expand a member to inspect their models and months.

The report includes successful calls with a positive user-billed cost for current active members whose email matches the organization's managed domain. Usage begins no earlier than both their organization membership and group membership. Departed and suspended members are excluded.

> **Note**
> **Allowance used** is usage for the selected report window, not the person's remaining balance or allowance reset window. A member in several groups can appear in each report. Do not add group totals together as an organization invoice or treat them as separate group bills.


## Find shared projects

Open **Shared projects** to see projects shared with the group that you can access. If **Load more projects** appears, more projects remain to be checked. An empty first page does not mean no matching projects exist.

Project owners manage access in the project's **Settings** → **Collaboration** panel. Open **Share** or **Manage**, find the group under **Groups**, and select the appropriate role. Group administration alone does not grant project ownership.

## If a change fails

If Bearly shows an error, confirm that you still administer the organization or group, then reopen **Groups** and check the current state before retrying. For an unavailable member, ask an organization admin to confirm that the person has an active organization membership. Change synced names and membership in the identity provider.

An organization admin can appoint another group admin if the group has no active administrator.

---

---
title: "Connect Microsoft Entra to an organization"
description: "Let a managed Bearly organization sign in with Microsoft work or school accounts, and provision members and groups from Microsoft Entra."
canonical: "https://bearly.ai/docs/teams/microsoft-entra/"
markdown: "https://bearly.ai/docs/teams/microsoft-entra.md"
---

# Connect Microsoft Entra to an organization

HTML: [https://bearly.ai/docs/teams/microsoft-entra/](https://bearly.ai/docs/teams/microsoft-entra/)

> Let a managed Bearly organization sign in with Microsoft work or school accounts, and provision members and groups from Microsoft Entra.

A managed organization can connect its Microsoft Entra tenant so that members sign in to Bearly with **Continue with Microsoft**. The same organization can also provision people and groups from Microsoft Entra, so joiners, leavers, and group changes reach Bearly without hand edits. Either part works on its own.

> **Note**
> The **Microsoft Entra** section appears only to administrators of a managed organization, which is an organization with a managed email domain. See [Request a managed organization](https://bearly.ai/docs/teams/overview.md#request-a-managed-team).


## Before you start

- You are an **Admin** of the managed organization.
- The organization's managed domain is a verified domain in your Microsoft Entra tenant, and each person's Microsoft account has an **Email** in that domain. Personal Microsoft accounts and guests from other organizations cannot use Microsoft sign-in.
- You have the tenant's **Tenant ID** from the Microsoft Entra admin center, under **Entra ID** → **Overview** → **Properties**.
- Someone who can grant tenant-wide admin consent in Microsoft Entra, such as a Global Administrator, Privileged Role Administrator, Cloud Application Administrator, or Application Administrator, is available to approve Bearly.

Bearly never asks for a Microsoft application secret. You provide only the tenant ID, and a Microsoft administrator approves Bearly on Microsoft's own consent page.

## Set up Microsoft sign-in



1. Open your account menu and follow **Organizations** → your organization → **Overview**. Scroll to **Microsoft Entra** and select **Setup Microsoft Entra**.

2. Under **Configure sign-in**, paste the tenant ID into **Directory tenant ID**, make sure **Allow Microsoft sign-in** is on, and select **Save and continue**.

   Bearly confirms with **Microsoft Entra login updated**, and the step shows **Saved**.

3. Under **Grant admin consent**, select **Connect Microsoft**. Microsoft's approval page opens. From the desktop app it opens in your system browser.

4. Sign in to Microsoft with an administrator account from the same tenant, review the permissions, and accept. Bearly asks only to sign people in and read their name and email address.

5. When Bearly shows **Microsoft connected**, select **Return to Bearly**. If you started from the desktop app, close the browser tab and return to the app.

6. Under **Test sign-in**, confirm the connection works: in another browser or after signing out, choose **Continue with Microsoft** with an account whose email ends in the organization's managed domain.



The **Microsoft sign-in** row now shows **Enabled**, and its button reads **Manage Microsoft Entra**. A tenant can be connected to one Bearly organization.

After approval, Bearly is listed under **Enterprise applications** in your Microsoft Entra tenant. Manage it there like any other application: require assignment under **Properties**, assign users and groups, or apply Conditional Access. Those Microsoft settings decide who can reach Bearly's sign-in. Bearly's own membership rules, described next, decide what happens after that.

## How members sign in

On the sign-in page, members select **Continue with Microsoft** and choose their work or school account. On the web, the sign-in continues in the same tab. In the desktop app, Microsoft opens in the system browser, and the app comes to the front and signs in when Microsoft finishes; the browser tab can then be closed. The iOS app does not offer Microsoft sign-in, so use email sign-in there.

Signing in with Microsoft uses the same Bearly account as signing in by email with that address, so existing chats and projects are unaffected. Email sign-in stays available to every member.

What happens on a person's first Microsoft sign-in depends on the organization's **Managed-domain access** setting under **Overview** → **Settings**:

- **Automatically add anyone with @domain**: the person becomes a member immediately and receives the organization's default policy.
- **Require an organization administrator to add each person**: the person must already be on the organization, either [added by an administrator](https://bearly.ai/docs/teams/members.md) or provisioned from Microsoft Entra as described below. Otherwise Bearly reports that the account is not allowed to join the organization.

A member whose **Account access** has been switched off cannot sign in with Microsoft until an administrator restores it.

## Turn off Microsoft sign-in

Select **Manage Microsoft Entra**, turn off **Allow Microsoft sign-in**, and select **Save and continue**. New Microsoft sign-ins for the tenant stop. Members keep their Bearly accounts and memberships and can continue with email sign-in.

## Provision members and groups from Microsoft Entra

Provisioning lets Microsoft Entra add members, keep their names current, suspend people who leave, and mirror groups into the organization. Bearly administrators keep control of roles, policies, and spending limits; provisioning never changes them.

### Generate the provisioning token in Bearly



1. Under **Microsoft Entra** on the organization's **Overview**, find **User and group provisioning** and select **Set up provisioning**.

2. Select **Generate token**.

3. Copy the **Tenant URL** and the **Secret token**. Bearly shows the token only once. If you lose it, generate a new one.



The row shows **Configured**, and its button reads **Manage provisioning**.

### Connect the Microsoft Entra provisioning service

Provisioning uses a dedicated enterprise application in your tenant. [Microsoft's SCIM tutorial](https://learn.microsoft.com/en-us/entra/identity/app-provisioning/use-scim-to-provision-users-and-groups#integrate-your-scim-endpoint-with-the-microsoft-entra-provisioning-service) has the full walkthrough; the steps below cover what Bearly needs.



1. In the Microsoft Entra admin center, go to **Entra ID** → **Enterprise apps** → **New application** → **Create your own application**. Name it, choose **Integrate any other application you don't find in the gallery**, and create it.

2. Open the application, select **Provisioning**, and start a new configuration. Enter Bearly's values in **Tenant URL** and **Secret Token**, select **Test Connection**, and save the configuration. If your admin center shows the older layout, choose **Get started**, set **Provisioning Mode** to **Automatic**, and enter the same values under **Admin Credentials**.

3. Under **Attribute mapping**, edit the user mappings to match the table below and confirm the group mappings. Save the changes.

4. Under **Users and groups**, assign the people and groups Bearly should receive. Keep the scope at **Sync only assigned users and groups**.

5. Select **Provision on-demand** and provision one user and one small group. Confirm they appear under **Members** and **Groups** in Bearly before going further.

6. Select **Start provisioning**. The first cycle can take a while. After that, Microsoft checks for changes about every 40 minutes.



#### User attribute mappings

| Microsoft Entra attribute | Target attribute | Notes |
| --- | --- | --- |
| `userPrincipalName` | `userName` | The matching attribute. It must be the person's email address in the managed domain. If user principal names differ from email addresses in your tenant, map `mail` instead, so provisioning and Microsoft sign-in refer to the same address. |
| `objectId` | `externalId` | Change this from the default `mailNickname`. |
| `Switch([IsSoftDeleted], ...)` | `active` | Keep the default expression. A person who is unassigned, disabled, or deleted in Microsoft Entra becomes suspended in Bearly. |
| `displayName` | `displayName` | Optional. Bearly uses it as the member's name. |

Bearly stores only these four attributes. It ignores any other user mapping, including given and family names, phone numbers, addresses, job title, department, manager, and roles, so you can leave Microsoft's default mappings in place. Roles are assigned in Bearly, never from Microsoft Entra.

For groups, keep the default mappings: `displayName` to `displayName`, `objectId` to `externalId`, and `members` to `members`. As with users, Bearly ignores any other group mapping.

### What provisioning changes in Bearly

- A newly assigned person is added as a **Member**. Someone already on the organization with the same email is linked rather than duplicated, and their role, policy, and limit stay as they are. If `displayName` is mapped, their name in Bearly updates to match.
- A person who is unassigned, disabled, or deleted in Microsoft Entra is suspended, not removed. They show **Suspended** under **Members**, lose access through the organization, and can no longer sign in with Microsoft, but their data is kept. Restore access by assigning them again in Microsoft Entra.
- Assigned groups appear under **Groups** with a **Synced** badge. While a group is syncing, its name and membership come from Microsoft Entra, so it cannot be renamed or deleted in Bearly. You can still assign a policy and priority and add extra members by hand; members that came from Microsoft Entra also show **Synced**.
- A group that Microsoft Entra stops provisioning becomes **Inactive** in Bearly. It keeps its policy, its project shares, and any members you added by hand, and it becomes active again if Microsoft Entra provisions it later.
- An **Inactive** synced group can be renamed or deleted, because Microsoft Entra no longer manages it. Do that when you retire a group in Microsoft Entra and create a replacement with the same name: until the old group gives up the name, Microsoft Entra cannot create the new one and reports a conflict every cycle.
- Changes made by provisioning appear under **Activity** attributed to **Identity provider**.

Limits to plan around:

- Every provisioned person needs an email address in the managed domain. Anyone else fails in Microsoft's provisioning logs.
- Group names must be 2 to 120 characters, and a single group update can include at most 500 members.
- Nested groups are not supported. Assign the groups whose direct members should sync.
- Provisioning cannot suspend the organization's last active administrator. Make another member an **Organization admin** first.

Manage provisioned people from Microsoft Entra rather than from **Members**. If you remove a provisioned member in Bearly while they are still assigned in Microsoft Entra, a later provisioning cycle can add them again.

### Rotate or disable the token

Select **Manage provisioning** under **Microsoft Entra**.

- **Rotate token** issues a new token and immediately stops the old one. Provisioning pauses until you paste the new token into the Microsoft Entra configuration and run **Test Connection** again.
- **Disable** stops provisioning and rejects the existing token. Existing members, groups, and memberships stay as they are, and Microsoft Entra can no longer update them. To start again, select **Generate token** and update the Microsoft Entra configuration.

## Troubleshooting

| What you see | What to do |
| --- | --- |
| **This Microsoft organization is not configured for Bearly.** | The account belongs to a tenant that is not connected, or **Allow Microsoft sign-in** is off. Check the saved **Directory tenant ID**. |
| **Your Microsoft account did not provide a work email.** | Ask a Microsoft administrator to set the person's **Email** in Microsoft Entra to an address in the managed domain. |
| **Microsoft did not confirm this account's email domain.** | The address uses a domain that is not verified in the tenant, for example an external address. Use an address in a verified domain. |
| **Use the Microsoft account for your Bearly organization.** | The account's email is outside the organization's managed domain. |
| **Your Microsoft account is valid, but it is not allowed to join this Bearly team.** | The organization requires an administrator to add each person, the membership is suspended, or that email is already linked to a different Microsoft account. An organization administrator can add or restore the member. |
| **This Microsoft sign-in expired or was already used.** | The sign-in must finish within a few minutes of starting. Start again from the sign-in page. |
| **Failed to update Microsoft Entra login** | Check that the tenant ID is complete and correct. A tenant can be connected to only one Bearly organization. |
| **Microsoft did not approve the connection.** | Consent was declined, or the account cannot grant admin consent. Select **Connect Microsoft** again and sign in with an eligible administrator. |
| **Microsoft tenant approval could not be verified.** | The approving account belongs to a different tenant than the saved **Directory tenant ID**. Correct the ID, or use an administrator of the saved tenant. |
| **Test Connection** fails in Microsoft Entra | The **Tenant URL** or **Secret Token** is wrong, or provisioning was rotated or disabled in Bearly. Generate a new token and paste both values again. |
| A provisioning log entry says the user name must be an email in the managed domain | The mapped attribute for that person is outside the organization's domain. Correct the person's address or the `userName` mapping. |

For help with a connected tenant, contact [Bearly Support](mailto:support@bearly.ai).

---

---
title: "Manage organization billing and seats"
description: "Start an organization billing account, buy and assign seats, change seat plans, and resolve payment issues."
canonical: "https://bearly.ai/docs/teams/billing/"
markdown: "https://bearly.ai/docs/teams/billing.md"
---

# Manage organization billing and seats

HTML: [https://bearly.ai/docs/teams/billing/](https://bearly.ai/docs/teams/billing/)

> Start an organization billing account, buy and assign seats, change seat plans, and resolve payment issues.

Organization billing lets an administrator buy Bearly seats for organization members. Each seat has its own Pro, Analyst, Research, or Max plan, while the organization uses one payment method, invoice history, billing interval, and renewal date.

Only organization administrators can open **Billing** or change seats. If the tab is not present, either you are an organization member or the organization's plan is provided through another arrangement. Ask an organization administrator or Bearly Support before starting a separate subscription.

## Start organization billing

The first purchase creates a billing account for the organization. It does not use, change, or cancel an administrator's personal subscription.



1. Open your account menu and follow **Organizations** → your organization → **Billing**.

2. Choose **Monthly** or **Yearly · 2 months free**. The first completed purchase fixes that interval for later seats on the organization account.

3. Select **Pro**, **Analyst**, **Research**, or **Max**, then set the number of seats.

4. Select **Start checkout** and complete the organization purchase in the billing page that opens.

5. Return to the organization's **Billing** tab. The **Subscription** card shows the account status, seat count, renewal date, and next invoice. New seats appear under **Seats** as ready to assign.



If you leave checkout before finishing, reopen **Billing** and retry the purchase. Bearly identifies an incomplete first checkout instead of treating it as an active subscription.

## Assign a seat

One teammate can hold one active organization seat.

1. Under **Seats**, find an open seat and select **Assign**.
2. Choose a teammate. Use **Search teammates…** to find someone by name or email when they are not already listed.
3. Select the teammate to assign the seat.

The seat row shows the teammate's email and plan after the assignment finishes. Someone who already holds a seat is marked **Already has a seat** and cannot be selected again.

## Add more seats

Select **Add seats**, choose the plan and quantity, then select **Add seats** again. All new seats use the billing interval fixed by the organization's first purchase. Bearly may open checkout or payment setup before the seats appear.

Open seats with the same plan are grouped together. Organizations with more than 20 active seats also get **Search seats…** for finding a holder, plan, or ending seat.

## Move or unassign a seat

Open the **Seat actions** menu beside an assigned seat:

- **Move seat** transfers the seat to another teammate. Confirming the move removes the previous holder's access from that organization seat immediately. The plan and organization billing date do not change.
- **Unassign** removes the current holder immediately and keeps the paid seat open for another teammate.

Moving or unassigning a seat does not change either person's personal Bearly subscription.

## Change a seat's plan

Open **Seat actions** and select **Change plan**. The confirmation explains when the change takes effect:

- An upgrade takes effect immediately. The holder receives the additional credits, and the organization is charged the prorated difference.
- A move to a lower plan is scheduled for the next renewal. The seat row shows the next plan until then.
- Selecting the current plan cancels a scheduled plan change.

Changing one seat does not change the other seats on the organization account.

## End or reinstate a seat

Select **End at renewal** from **Seat actions** to stop a seat from renewing. The holder keeps the seat through the current billing period, and the row shows its end date.

Before that date, open **Seat actions** and select **Reinstate** to keep the seat active and renewing.

## Manage payment and invoices

The **Subscription** card includes **Payment method** and **Invoices**. If the account becomes past due, select **Fix payment**. Organization-seat access is blocked until the open invoice is paid, although a holder may still have access from a separate personal plan.

> **Note — Personal plans remain separate**
> Bearly uses the higher of a person's active personal plan and organization seat; their allowances are not added together. Nothing is canceled automatically. When an organization seat makes an eligible web-purchased personal plan redundant, the holder can open **Settings** → **Billing** and select **End personal plan** to schedule that subscription to end at its own period boundary. App Store subscriptions remain managed through the iOS app.


## If a seat action is unavailable

| What you see | What to do |
| --- | --- |
| No **Billing** tab | Confirm that you are an organization administrator. If the organization already receives a plan through another arrangement, contact Bearly Support instead of starting self-serve billing. |
| A teammate does not appear | Confirm that they are listed under **Members** and do not already hold an active organization seat. |
| A different monthly or yearly interval is unavailable | The organization's first completed seat purchase fixed the interval for the account. |
| **Payment is past due** | Select **Fix payment**. Seat access returns after the open invoice is paid. |
| **Unable to load organization billing** | Check your connection and reopen **Billing**. If the message continues, contact `support@bearly.ai` before retrying a purchase. |

For personal subscriptions and usage, see [Manage billing and review usage](https://bearly.ai/docs/account/billing-usage.md).

---

---
title: "Set organization policies"
description: "Set shared chat defaults, model access, feature permissions, and spending limits for members of a managed organization."
canonical: "https://bearly.ai/docs/teams/policies/"
markdown: "https://bearly.ai/docs/teams/policies.md"
---

# Set organization policies

HTML: [https://bearly.ai/docs/teams/policies/](https://bearly.ai/docs/teams/policies/)

> Set shared chat defaults, model access, feature permissions, and spending limits for members of a managed organization.

Policies give an organization administrator one place to set the defaults, tools, models, and limits available to groups of members.

> **Note**
> Policies are available to administrators of managed organizations. Members do not see the **Policies** tab. On an organization that is not managed, the tab shows **Managed only** and an **Upgrade to Managed** option instead of the policy editor.


## Open your organization's policies



1. Open your account menu and select **Organizations**.
2. Select the organization you want to manage.
3. Select **Policies**.



The page lists each policy and marks the organization's **Default** policy. If the organization has no policies yet, the first one you create becomes the default for new members.

## Create a policy



1. Select **Add policy**.
2. Enter a **Policy name** between 4 and 18 characters.
3. Optionally enter a **Monthly spend limit**. Leave it blank for no policy-level limit.
4. Configure the chat defaults, model access, and feature controls your members need.
5. Review the footer for the **Unsaved changes** indicator, then select **Save**.



The monthly limit applies to managed-domain members assigned to the policy. A limit set directly on an individual member overrides the policy limit for that person.

### Set chat defaults

Turn on **Enable chat defaults** to choose the **Model**, **Instructions**, **Language**, and **Temperature** that new chats outside projects start with. Project chats use their project settings instead of these policy chat defaults.

Each default can be **Unlocked** or **Locked**:

- **Unlocked** gives members the default while allowing them to change it.
- **Locked** prevents members from changing that setting.

Temperature runs from more focused to more creative. The **Factual**, **Default**, and **Creative** presets provide useful starting points.

### Control model access

Under **Models**, choose one of three **Model access** modes:

- **Off** leaves chat models unrestricted by this policy.
- **Allow-list** permits only the models you add to **Allowed models**.
- **Deny-list** permits models except those you add to **Denied models**.

Use **Filter available models** to find a model, then select **Add** or **Remove**. An allow-list must contain at least one model before the policy can be saved.

A **ZDR** badge marks each model whose provider does not retain prompts or responses after processing them. Search for `zdr` in the filter to narrow the list to those models. See [Use zero-data-retention models](https://bearly.ai/docs/zero-data-retention.md).

Model access controls chat models. Use the feature sections below it to permit or restrict Bearly's other tools.

### Choose which features members can use

Under **Feature access**, choose how the policy handles features Bearly adds later:

- **Enabled by default** makes new features available automatically. New policies use this setting.
- **Disabled by default** keeps new features unavailable until you switch them on.

Changing this default keeps access to every current feature unchanged. It only determines the starting state of features added later.

Each feature has a standard switch. On means members can use the feature, or that the named requirement is enforced. Off means members cannot use it. The controls are arranged by the kind of work they affect:

| Section | What you can manage |
| --- | --- |
| **Web & Research** | Web tools, web search, Image Search, Deep Research, People Search, X, SEO data, Domain research, Maps, and Weather |
| **Media & Creativity** | Images, video, Pages, Charts, voice, transcription, and text to speech |
| **Code & Automation** | **Run Python**, local work, Computer Use, Browser, Connectors, personal MCP servers, chat management, email notifications, text messages, Routines, and browser use in Routines |
| **Collaboration** | Sharing, public Page links, sharing Pages with people and teams, Screen Share, Council, and Meetings |
| **Privacy & Security** | Memory Recall, the PII Blocker, local encryption, existing-chat locking, and file uploads |
| **Skills** | Whether members can use, add, and run Skills, including Skills supplied by a policy |
| **Projects** | Project creation, password requirements, collaboration, and project documents and data |
| **Developer** | Advanced developer access when it is available to your organization |

The exact controls shown can depend on the features available to your organization.

Full access eligibility is managed separately under **Overview** → **Agent access**. Only organization administrators can change its default or individual exceptions; group administrators and ordinary feature policies cannot grant it. Members must then enable Full access for each chat or separately for each routine on its device. Allowing Full access does not turn on tools disabled by a policy.

Turning off the Full access default leaves individual allowances in place. To remove eligibility for everyone, use **Turn off for everyone** under **Agent access**. This removes individual exceptions while preserving saved tool approvals and routines' separately enabled unattended permissions. See [Choose how Bearly uses external tools](https://bearly.ai/docs/settings/agent-permissions.md#manage-full-access-for-an-organization) for the controls, confirmation and recovery steps.

Under **Privacy & Security**, **Unitemized billing** lets members choose daily totals for their own accounts.

Under **Code & Automation**, **Browser in Routines** controls whether members can let a scheduled routine use Bearly Browser while they're away, on the sites they approve and without asking first, including clicking, typing, and submitting forms when they allow it. A routine also needs **Browser** on, and the switch is unavailable while **Routines** is off. Turning it off doesn't affect Browser or Chrome in chats. See [Let a routine use a browser](https://bearly.ai/docs/features/routines.md#let-a-routine-use-a-browser).

Under **Collaboration**, **Sharing** is the master control for Page publishing. **Public links** permits **Anyone with the link**, while **Share with people and teams** permits restricted links for selected members and teams. To allow restricted Page links without public links, turn on **Sharing** and **Share with people and teams**, then turn off **Public links**. These controls do not disable the owner's live Page preview.

> **Note**
> Some features depend on another feature. For example, Deep Research requires **Web** and **Web search**. A dependent switch is unavailable while its requirement is off.


> **Danger — Locking existing chats cannot be undone**
> **Lock Existing Chats** restricts affected members' existing chats to local access. The keys are unrecoverable, and the confirmation cannot be reversed. Make sure your organization understands the effect before enabling it.


### Provide MCP servers to members

Under **Code & Automation**, **MCP Connectors** controls the whole Connectors feature. **Personal MCP servers** controls whether members can add their own custom servers; turning it off does not remove Bearly's built-in services or servers supplied by the policy.

Use **Team MCP servers** to give every member on this policy the same server definition:

1. Select **Add server**.
2. Choose **HTTPS** or **npm package**.
3. Enter the server name and connection details. For bearer tokens, custom headers, or npm environment variables, enter the required variable names only.
4. Select **Add server**, then save the policy.

The server appears automatically in each assigned member's **App Settings** → **Connectors**. Members cannot change or remove its definition. They connect with their own OAuth account or enter their own token, header, or environment values. Those values never become part of the policy.

HTTPS servers work in desktop and mobile browsers and in the desktop app. In the iOS and Android apps, custom HTTPS servers can use a bearer token, custom headers, or no authentication. Services requiring OAuth sign-in must be connected and used in a regular browser; that connection does not carry back to the mobile app. npm servers require the desktop app. A policy does not remove these platform limits.

See [Connect services and MCP servers](https://bearly.ai/docs/features/connectors.md) for the member setup and runtime behavior.

### Disable new features by default for every policy

On the policy list, select **Disable new features by default** to update every policy where new features currently start enabled. Confirm **Disable by default**.

Bearly preserves each policy's current feature access. Features added later start unavailable until an administrator enables them. If a policy cannot be updated, Bearly names it so you can retry the same action safely.

## Make a policy the default

The default policy is assigned automatically to new managed-domain members. To change it:

1. Open the policy you want to use.
2. Select **Make Default**.
3. Confirm **Make Default**.

Changing the default does not reassign existing members. To change an existing member, open **Members**, select the member, choose a **Policy**, and select **Save**.

## Review or share a policy summary

Open an existing policy and select **Summarize the Policy** to generate a plain-language summary. If you have unsaved edits, the summary identifies that it reflects those edits.

Select **Email me a copy** to send the current summary to your account email. If Bearly cannot generate the summary, select **Try again**. An organization's model restrictions can prevent summary generation when no suitable model is available.

## Edit or delete a policy

Select a policy from the list to edit it. Bearly marks pending edits as **Unsaved changes**; select **Save** to apply them or **Cancel** to leave the editor.

Only a non-default policy offers **Delete**. Deleting a policy does not automatically move its current members to another policy: they continue using the deleted policy's settings until an administrator changes their assignment, and the policy can appear as **Inactive** beside those members. Reassign them before deleting the policy if you want a clean transition.

## If a policy will not save

- Check that **Policy name** contains 4–18 characters.
- If **Model access** is set to **Allow-list**, add at least one model or switch the mode to **Off**.
- If Bearly shows **Failed to save policy. Please try again.**, keep the editor open and select **Save** again after checking your connection.

---

---
title: "Review organization usage"
description: "Compare a managed organization's AI spend and calls across members, models, and recent months."
canonical: "https://bearly.ai/docs/teams/usage/"
markdown: "https://bearly.ai/docs/teams/usage.md"
---

# Review organization usage

HTML: [https://bearly.ai/docs/teams/usage/](https://bearly.ai/docs/teams/usage/)

> Compare a managed organization's AI spend and calls across members, models, and recent months.

> **Note — Limited availability**
> Organization **Usage** is available to administrators of eligible managed organizations. If the tab is not present, the feature has not been enabled for that organization or your role does not include access.


The organization Usage view answers two different questions: how much the organization used overall, and which members and models account for that activity. It reports AI calls and their cost; it is separate from an individual member's credit meter and personal **App Settings** → **Usage** page.

## Open the organization report



1. Open your account menu and select **Organizations**.
2. Select the managed organization you administer.
3. Select **Usage** in the organization navigation.
4. Choose **1 mo**, **3 mo**, **6 mo**, or **12 mo** to set the reporting window.



The summary cards update to show:

- **Spend** across the selected window.
- **Calls** made by members in that window.
- **Active members**, shown against the total number of members in the report.
- **Top model**, based on the organization's usage during the window.

Changing the window reloads the report and closes any expanded member row.

## Understand member usage

Under **Members**, each row shows the member's name and email, total calls, cost, and most-used model. Select a member to expand two breakdowns:

- **Models** lists that member's calls and cost for each model.
- **Months** divides the same activity by calendar month.

Use **Search members** to filter by name or email. A member with no activity in the selected period remains visible with **No usage**, making it easier to distinguish an inactive account from a missing one.

> **Caution**
> Cost is useful for administration, not as a measure of productivity or work quality. A few demanding research tasks can cost more than many routine messages, and the report does not explain the content or value of a member's work.


## If the report is incomplete

- If **Usage** is missing, confirm that you are an administrator of a managed organization and ask Bearly Support whether the report is available for the organization.
- If the page says **Could not load usage.**, check the connection, leave the tab, and open it again.
- If search says **No members match your search.**, clear or shorten **Search members**.
- If totals look unexpectedly small, check the selected month window and expand the relevant member to compare the **Models** and **Months** rows.
- For an individual allowance, reset date, or credit boost, use [Manage billing and review usage](https://bearly.ai/docs/account/billing-usage.md).

---

---
title: "Distribute skills from a shared folder"
description: "Register a managed organization's shared skill library and assign individual skills through organization policies."
canonical: "https://bearly.ai/docs/teams/skill-stores/"
markdown: "https://bearly.ai/docs/teams/skill-stores.md"
---

# Distribute skills from a shared folder

HTML: [https://bearly.ai/docs/teams/skill-stores/](https://bearly.ai/docs/teams/skill-stores/)

> Register a managed organization's shared skill library and assign individual skills through organization policies.

> **Note — Limited availability**
> **Skill stores** are available to administrators of eligible managed organizations. Registration and assignment require the Bearly desktop app. If the organization has no **Skills** tab, the feature is not available for that organization yet.


A skill store lets a managed organization distribute approved Bearly skills from a shared folder. Administrators register the folder once, inspect the skills Bearly finds, and enable each skill on one or more organization policies.

Members assigned to those policies must also be able to reach the shared folder from their computers. Before rollout, agree on a mounted location that is consistent for the organization.

## Prepare the organization

- Create at least one [organization policy](https://bearly.ai/docs/teams/policies.md). Skills are assigned through policies, not directly to individual members.
- Put each ready-to-use skill in its own folder within the shared location. Bearly lists folders that contain a skill definition and groups them by their top-level folder.
- Mount the shared drive on the administrator's computer and on every member computer that should use the skills.

## Add a skill store



1. Open Bearly desktop, then select your account menu → **Organizations**.
2. Select the managed organization and open **Skills**. The page heading is **Skill stores**.
3. Under **Add a skill store**, select **Shared folder** → **Choose folder…**, then choose the mounted folder.
4. Enter a recognizable **Store name**, such as `Editorial skills`.
5. Select **Add Store**.



Bearly confirms that the store was added, opens it, and reads the available skills. Register the shared root rather than a private copy on one person's computer.

## Assign skills to policies



1. Select the store to expand it. Bearly shows **Reading skills…** while it checks the folder.
2. Open a folder group, then review each skill's name and description. A **Python** badge means the skill includes scripts and deserves the same review you would give other executable work.
3. Under **Enable on Policies**, select a policy name to assign the skill. A checkmark marks an active assignment.
4. Select the same policy again to remove that assignment.



The skill row reports **Unassigned** or how many policies currently use it. Policy members receive the assignment when Bearly can reach that shared store on their computer.

> **Caution — Review shared skills before assigning them**
> A skill can shape how Bearly works and may include scripts or supporting files. Treat the shared folder as managed software: control who can change it, review updates, and test a skill with a limited policy before broad assignment.


## Remove a store

Select **Remove store** on the store row. Removing the registration also removes that store's assignments from the organization's policies. It does not delete the shared folder or its files from the drive.

If the folder is only temporarily unavailable, leave the store registered and restore the drive connection instead of removing it.

## If Bearly cannot find the skills

| What you see | What to do |
| --- | --- |
| **Create an organization policy first — skills are assigned to policies.** | Create a policy, then return to **Skills**. |
| **Could not read this store** | Open Bearly desktop on a computer where the shared drive is mounted, confirm that the folder still uses the registered path, then open the store again. |
| **No SKILL.md folders found under this path.** | Check that the selected root contains properly prepared skill folders, rather than only documents or an extra empty nesting folder. |
| A member does not receive an assigned skill | Confirm the member's policy assignment, that the policy is checked on the skill, and that the same shared path is mounted on the member's computer. |
| **Skills** is missing | Confirm that you are a managed organization administrator. If you are, the limited feature is not available for the organization yet. |

For creating and using personal skills, see [Use skills in Bearly](https://bearly.ai/docs/features/skills.md).

---

---
title: "Review organization activity"
description: "See who changed your organization's settings, policies, and membership, with readable before-and-after details."
canonical: "https://bearly.ai/docs/teams/activity/"
markdown: "https://bearly.ai/docs/teams/activity.md"
---

# Review organization activity

HTML: [https://bearly.ai/docs/teams/activity/](https://bearly.ai/docs/teams/activity/)

> See who changed your organization's settings, policies, and membership, with readable before-and-after details.

**Activity** gives organization administrators a chronological record of important changes to an organization. Use it to understand what changed, who made the change, and which member, policy, or organization setting was affected.

> **Note**
> Only organization administrators can see **Activity**. It is available for both managed and unmanaged organizations.


## Open Activity



1. Open your account menu and select **Organizations**.
2. Select the organization you want to review.
3. Select **Activity**.



The newest changes appear first. Each entry includes:

- A short description of the change
- An **Organization**, **Settings**, **Policy**, **Member**, or **Activity** category
- When the change happened
- The email address of the person who made it
- The affected member, policy, or organization setting when one is available

On a computer, hover over a relative time, such as “3 hours ago,” to see its full date and time.

## Inspect what changed

Select an entry with an expand arrow to see its details. Bearly presents changed fields as a before-and-after comparison—for example, a member role changing from **Member** to **Admin**, or a policy setting changing from **Allowed** to **Disabled**.

Some grouped settings are summarized as **Previous settings** and **Changed settings** rather than displaying every nested value.

## Changes included in Activity

Activity records these organization-management events:

- Organization creation
- Organization name and default-policy changes
- Policy creation, editing, and deletion
- Member additions, edits, and removals
- Changes to member roles, names, assigned policies, and spending limits

Activity is a history of organization administration. It does not list chats, prompts, model calls, project work, or general member usage.

> **Caution**
> Activity begins when change recording became available; earlier organization changes were not added retroactively. An older organization can therefore have a shorter history than its age suggests.


## See older entries

Select **Load more** at the end of the list to retrieve the next set of older changes. Continue until **Load more** no longer appears.

If no entries are available, Bearly shows **No organization changes recorded yet.** This can mean that no supported organization changes have happened since Activity began recording them.

## If Activity does not load

If Bearly shows an error, select **Retry**. If the problem continues, confirm that you are still an organization administrator, then reopen **Organizations** → your organization → **Activity**.

---

---
title: "Change your chat settings"
description: "Choose how Bearly responds in one conversation, set defaults for new chats, and control the personal context shared with AI models."
canonical: "https://bearly.ai/docs/settings/chat-settings/"
markdown: "https://bearly.ai/docs/settings/chat-settings.md"
---

# Change your chat settings

HTML: [https://bearly.ai/docs/settings/chat-settings/](https://bearly.ai/docs/settings/chat-settings/)

> Choose how Bearly responds in one conversation, set defaults for new chats, and control the personal context shared with AI models.

Chat settings let you choose the model, language, response style, memory, web search, reasoning, instructions, and creativity that Bearly uses. You can change one conversation without affecting the rest, or establish defaults for chats you create later.

The same Settings window is available in Bearly chat on the web, desktop, iOS, and Android. A feature-specific control may be absent on a platform that does not support that feature.

## Open chat settings

From an open conversation:



1. Select the sliders icon beside the message box.
2. Under **Settings**, select **Chat Settings**.
3. Bearly opens **Settings** → **Chats** → **This Chat**.



You can also open your account menu and select **App Settings**. Choose **This Chat**, **New Chats**, or **Personalization** under **Chats**. On a narrow screen, these panes appear in the selector at the top of Settings instead of a list beside the content.

## Choose the right scope

| Pane | What it changes |
| --- | --- |
| **This Chat** | The open conversation only. It does not change another existing chat or your defaults. |
| **New Chats** | Chats you create afterward. It does not rewrite settings in existing conversations. |
| **Personalization** | Profile context that can be included across your conversations when you turn it on. |

In an ordinary chat, **Reset all to defaults** appears after you override one or more **This Chat** settings. Select it to return that conversation to your **New Chats** choices.

> **Note — Projects have their own defaults**
> **New Chats** does not provide general defaults while you are inside a project. Open **Project Settings** to configure project chat settings. A project's settings take priority over your general **New Chats** choices.


## Choose how Bearly responds

The controls shown in **This Chat** and **New Chats** are nearly identical:

| Setting | What it controls |
| --- | --- |
| **Model** | The AI model used for the conversation. **Recommended**, available under **New Chats**, lets Bearly choose an available model rather than pinning one. |
| **Language** | The language Bearly should use for its responses. |
| **Response Style** | The usual tone and level of detail. Choose **Normal**, **Classic**, **No Style**, **Learning**, **Concise**, **Explanatory**, or **Formal**. |
| **Memories** | **Auto** lets Bearly recall relevant context from earlier chats; **Off** prevents automatic memory recall in this chat. |
| **Web Search** | **Auto** lets Bearly decide when current web information would help; **Off** prevents automatic web search. |
| **Reasoning Effort** | Choose **Default**, **None**, **Low**, **Medium**, **High**, or **Extra High** to guide how much reasoning the selected model should apply. |
| **Instructions** | Add standing directions for the conversation, such as the audience, preferred format, or facts Bearly should keep in mind. |
| **Temperature** | Move from **Focused** toward **Creative**, or use **Factual**, **Default**, and **Creative** presets. Values above 1.5 may produce incoherent output. |

Available models and controls can differ by plan, platform, project, and team policy. If an administrator has locked a setting, Bearly uses the managed value and may omit that control from **This Chat**.

## Manage long-conversation history

History is a conversation-level control in the chat menu rather than the Settings window:

1. Select the sliders icon beside the message box.
2. Under **Search**, select **History**.
3. Choose one of the two modes:

   - **Chat (Dynamic)** is the default. It keeps the conversation moving by reducing older context as needed.
   - **Agentic (Compaction)** summarizes older parts of a long conversation when necessary. When that happens, the chat shows **Conversation compacted so you can keep going**.

Changing **History** affects the open conversation. It does not change the history mode of other chats.

## Add personal context

Open **App Settings** → **Chats** → **Personalization**, then:



1. Turn on **Personalization**.
2. Add the **Name** Bearly should use, your **Occupation**, and any useful context under **About you**.
3. Wait for **Saved** before closing Settings.



When Personalization is on, the profile details you enter are shared with the AI models used in your conversations so they can tailor their responses. Turning Personalization off stops including that profile context; the fields remain unavailable until you turn it on again.

A project owner can disable Personalization for a project. In that case, Bearly does not include your profile context in that project's chats even if the account-level switch is on.

## Know when a change is saved

Settings save automatically. The Settings header can show **Saving...**, **Saved**, or **Not saved**.

- If you see **Saving...**, leave Settings open until the save completes.
- If you see **Not saved**, check your connection and make the change again. Settings remains open when a pending change cannot be saved.
- For **Instructions** and other text fields, click or tap outside the field after editing, then wait for **Saved**.

## If a setting is missing or has no effect

| What you see | What to do |
| --- | --- |
| **New Chats** says defaults are unavailable in project chats | Open that project's **Project Settings** and change its chat settings there. |
| A model or control is not listed | It may be unavailable on your plan, restricted by a team policy, or disabled for the project. Ask the project owner or team administrator when applicable. |
| One conversation behaves differently from your defaults | Open **This Chat** and select **Reset all to defaults** if the option is shown. |
| Responses are erratic or incoherent | Lower **Temperature** to **Default** or **Factual**, then send the request again. |
| Personalization does not affect a project chat | Check whether the project's settings disable Personalization. |
| Settings shows **Not saved** | Keep Settings open, restore your connection, and retry the change. |

For more on memory scope in projects, see [Use memory and search](https://bearly.ai/docs/features/memory-search.md).

---

---
title: "Choose how Bearly uses external tools"
description: "Choose Standard or Full access for chats and routines, review saved access, and manage organization eligibility."
canonical: "https://bearly.ai/docs/settings/agent-permissions/"
markdown: "https://bearly.ai/docs/settings/agent-permissions.md"
---

# Choose how Bearly uses external tools

HTML: [https://bearly.ai/docs/settings/agent-permissions/](https://bearly.ai/docs/settings/agent-permissions/)

> Choose Standard or Full access for chats and routines, review saved access, and manage organization eligibility.

Use **Permissions** beside the chat input to choose when Bearly asks before using enabled tools outside the chat.

**Standard** asks for approval when needed and respects approvals you have already saved. **Full access** lets Bearly use enabled external tools without asking each time. The selected model does not change this permission mode.

> **Note — Limited availability**
> Full access must be available for your account and app version. For a managed organization account, an organization administrator must allow it first. Each tool still needs to be available on your device: local commands, Computer Use, and Bearly Browser or connected Chrome control require the desktop app.
>
> On a phone or tablet, Full access applies to supported [connectors](https://bearly.ai/docs/features/connectors.md) enabled for that chat. Mobile browsers support HTTPS connectors, including service sign-in. In the iOS and Android apps, custom HTTPS servers can use a bearer token, custom headers, or no authentication; services requiring OAuth sign-in need Bearly in a regular browser. Full access on your phone does not control another device.


## Enable full access for a chat



1. Open the chat and select **Permissions** beside the input.
2. Select **Full access**.
3. Review the enabled tools and warning, then select **Enable full access**. Choose **Cancel** to keep Standard.



The composer shows **Full access** while it is active. It applies only to this chat on this device. New chats, shared copies, another account, and other devices do not inherit it.

> **Caution — Review the access you are giving**
> Enabled tools may read, change, or delete files, run programs, and act in signed-in apps and websites. Mistakes or instructions in untrusted content can cause unintended changes. Keep sensitive work in Standard when you want to review actions individually. A selected working folder is context, not a security sandbox.


Organization restrictions, operating-system permissions, and required safety checks still apply. Full access does not turn on a disabled tool, connect an account, or grant access that the account does not have.

Tools and accounts you deliberately enable later are included. Their setup screens explain that chats using Full access can use them without asking for tool approval. Scheduled [Routines](https://bearly.ai/docs/features/routines.md) require their own separate Full access approval.

## Return to a chat

Restarting Bearly or returning after 24 hours away pauses Full access. The composer shows **Full access paused**. Select **Resume** to use it again after Bearly checks availability, or **Use Standard** to turn it off.

While paused, the chat follows Standard. Sending a message does not resume Full access. If your administrator has removed eligibility, Resume remains unavailable.

If Bearly cannot check availability, reconnect and open **Permissions** or **Full access paused** again. If the panel asks for an app update or an enabled external tool, complete that step first. An organization restriction needs your administrator's help; it cannot be changed from the chat.

## Return to Standard or remove saved approvals

Select **Full access** beside the input, then **Standard**. Bearly withdraws Full access and stops current work where it can. Actions already completed outside Bearly are not undone.

Changing modes keeps your earlier saved approvals. Review them under **Saved approvals** in the same panel. In Standard, select **Revoke** beside one, or **Revoke all saved approvals**, to stop current work and ask again for those actions. **Clear all permissions** switches to Standard and removes every saved approval together. Stop by itself ends the current task without changing your permission mode.

## Choose full access for a routine

A routine does not inherit Full access from its chat. Save the routine first, then use **Tools** → **Permission mode** in its editor on the desktop device that runs it. Select **Enable full access** and review the warning and selected tools before confirming. Enabling it stops any current run and leaves the schedule unchanged.

Routine consent stays on across app restarts and time away. Each run still needs current authorization and can use only the supported tools and accounts selected for that routine. Changes to its saved instructions, schedule, model, delivery settings, tools, or selected connector and skill setup require review. **Full access needs review** means the routine cannot continue in that mode until you approve its current setup. If its schedule has paused, select **Resume** after reviewing. Copying or moving a routine starts it in Standard.

**Use Standard** withdraws routine Full access and stops current work on its running device while keeping existing per-tool unattended permissions, including local or browser access you enabled separately. Standard routines can still use those permissions without asking. **Stop run** keeps the mode and schedule. A change made from web, mobile or another computer applies when it syncs to the running device. Those other devices can manage the routine, but Full access must be enabled on its installed desktop device.

Routines cannot use Computer Use, Google Workspace or Chrome DevTools in either mode. Their browser tools keep their own tabs and stay on the sites and actions approved for the routine. See [Run prompts on a schedule with Routines](https://bearly.ai/docs/features/routines.md#choose-the-routines-permission-mode) for the complete setup and limits.

## Manage full access for an organization

Organization administrators manage eligibility for members on the organization's managed domain. Group administrators cannot grant Full access.



1. Open the account menu, select **Organizations**, then select your organization.
2. On **Overview**, find **Agent access**.
3. Set **Allow full access by default**. This starts off for existing organizations.
4. To set an exception, open **Members**, select the member's **Member settings** gear, and choose **Use organization default**, **Allow full access**, or **Do not allow full access** under **Full access**. This setting saves immediately.



The default applies to members using **Use organization default**. Individual **Allow full access** or **Do not allow full access** choices take precedence, so turning off the default alone does not remove individual allowances.

An allowed member still chooses whether to enable Full access for a chat or separately for a routine. This setting is separate from ordinary feature policies; allowing Full access does not enable tools that your organization has disabled.

To withdraw eligibility across the organization, select **Turn off for everyone** under **Agent access**. Review the current number of eligible members, then confirm. This turns off the default and removes individual exceptions. Full access authorization for active chats and routines ends within 60 seconds; saved approvals for individual tools remain. Previously authorized external actions may already have taken effect. Routines require a new review before using Full access again. Changes appear in organization **Activity**.

If settings cannot load or another administrator changed them, select **Reload** before trying again. In the confirmation dialog, use **Reload affected members** to check the count again. An unavailable setting does not grant access.

For the tools themselves, see [Work locally](https://bearly.ai/docs/features/work-locally.md), [Computer Use](https://bearly.ai/docs/features/computer-use.md), [Browser](https://bearly.ai/docs/features/browser.md), and [Connectors](https://bearly.ai/docs/features/connectors.md).

---

---
title: "Manage billing and review usage"
description: "Change your Bearly plan, review the current credit period, add a credit boost, and see monthly usage by model."
canonical: "https://bearly.ai/docs/account/billing-usage/"
markdown: "https://bearly.ai/docs/account/billing-usage.md"
---

# Manage billing and review usage

HTML: [https://bearly.ai/docs/account/billing-usage/](https://bearly.ai/docs/account/billing-usage/)

> Change your Bearly plan, review the current credit period, add a credit boost, and see monthly usage by model.

Billing settings show your current plan and the actions available for that subscription. Paid accounts also get a personal Usage view with the current credit period and a month-by-month model breakdown.

For a short support-ready cancellation procedure, see [Cancel your Bearly subscription](https://bearly.ai/docs/account/cancel-subscription.md).

## Open Billing or Usage



1. Open the account menu that shows your email, plan badge, and usage meter.
2. Select **Billing & Pro** to open **Settings** → **Billing**.
3. To review a paid account's activity, select **Usage** under **Settings**.



You can also select **App Settings**, then choose **Billing** or **Usage**. The account menu's meter gives you a quick percentage at any time; **+boost** appears beside it when boost credits are active.

> **Note**
> **Billing** is available on free and paid accounts. **Usage** is available on paid plans; free accounts can use the account-menu meter to track their daily allowance.


## Review or change your plan

The **Current Plan** card identifies your plan and shows the actions available for its payment source. Depending on the account, these can include **Manage Account**, **Invoices**, **Restore**, and **Cancel**.

To compare plans:

1. Choose **Monthly** or **Yearly**. The yearly option is marked **17% off**.
2. Review the **Pro**, **Analyst**, and **Research** cards. If you already have a paid plan, the **Free** card is also shown.
3. Select **Upgrade to _Plan_**, **Switch to _Plan_**, or **Purchase _Plan_**, as shown.
4. Complete the purchase or account change in the billing page that opens.

Your active choice is marked **CURRENT** and cannot be selected again. If a cancellation is scheduled, **Current Plan** shows the date under **Cancels on** and offers **Reactivate**.

> **Caution — Use the account that purchased the plan**
> Subscriptions belong to the Bearly account and payment account used for the purchase. If **Restore** finds no subscription, confirm that you signed in with the correct Bearly account before purchasing again.


## Understand credits and reset periods

- Free accounts use a daily allowance. The account-menu meter shows the percentage used, and the allowance resets daily.
- Paid plans use a monthly allowance. Open **Usage** to see **Used this month**, **Monthly limit**, today's usage, and the displayed reset date.
- A managed team or individual member limit can change the monthly limit shown for that account.
- Credit use varies by model and task. The percentage meter is a better guide than counting messages.

If a chat reaches its limit, select **View Billing**, **Buy Boost**, or **Boost or Upgrade** in the message Bearly shows. The available action depends on your plan, payment source, and boost eligibility. After adding credits or changing plans, return to the chat and send the request again.

## Add a Credit Boost

Eligible paid, self-managed accounts can add one-time credits under **Billing** → **Credit Boost**. Boosts are used during the current credit cycle and expire with that cycle.

The card shows the choices available to your account, the credits added, and the amount you will pay. Bearly may also offer **Claim _amount_ Credits** when a free boost is available.



1. Review **Credit Boost**, including any **Boost remaining** balance.
2. Select a credit amount and check the **Pay _amount_** price before continuing.
3. Complete checkout, then return to Bearly and reopen **Billing** or **Usage** to confirm the new balance.



Repeated boosts can use capacity pricing, which the card identifies before purchase. Each plan also has an outstanding-credit limit. When the limit is reached, Bearly hides unavailable options and explains that the plan's cap has been reached. If boosts are becoming routine, the card may show a less expensive plan upgrade instead.

Boosts are not offered to free accounts, team-managed billing, or inside the iOS app. A paid App Store subscriber can open Bearly outside the iOS app to buy an available one-time boost, while plan changes continue through Apple.

## Review personal Usage

Open **App Settings** → **Usage**. The **Current Period** card shows:

- Your plan and the percentage used.
- When credits reset.
- The amount used and your current limit.
- Today's cost on a monthly plan.
- **Boost remaining**, when applicable.

Select **Manage** to return to Billing. When monthly credits apply, **Buy Boost** opens the same Billing pane.

Under **Usage by service**, use **Previous month** and **Next month** to move between available months. The table shows **Service**, **Calls**, and **Cost**, followed by a **Total** row. You can go back to the month you joined Bearly and forward only as far as the current month.

With **Unitemized billing** on, usage appears as one **Bearly Private** line with no call count. See [Turn on unitemized billing](https://bearly.ai/docs/unitemized-billing.md).

If there is no recorded activity for the selected period, the table is replaced by **No usage this month.**

> **Note**
> This is your personal AI usage. Team administrators may have a separate team Usage view for comparing members, depending on the team's access.


## Payment and account differences

| Account | What to expect |
| --- | --- |
| Purchased on the web | Plan changes and account management open a secure browser billing page. |
| Purchased in the iOS app | Manage the subscription through the iOS app. Outside iOS, plan cards show **Manage in iOS App** and cannot be selected. |
| Team-managed | **Billing** shows **Team-Managed Billing** instead of plan cards. Ask your team administrator or select **Contact Support** for billing help. |

## If Billing or Usage looks wrong

| What you see | What to do |
| --- | --- |
| **Unable to redirect** or the billing page does not open | Check your connection, then select the action again. If it continues, contact `support@bearly.ai`. |
| **Checking for previous subscription...** does not resolve | Confirm the purchasing account, reopen Billing, and select **Restore** again. |
| An iOS purchase does not appear immediately | Leave Bearly open briefly while it checks the subscription, then reopen **Billing**. Use **Restore** if needed. |
| A boost says it is no longer available or would pass the plan limit | Review the refreshed options and choose one that remains available. |
| **Usage by service** unexpectedly says **No usage this month.** | Confirm the selected month, check your connection, and reopen **Usage**. |
| Your account says **Team-Managed Billing** | Ask the team administrator about the assigned plan or limit; use **Contact Support** for payment assistance. |

---

---
title: "Cancel your Bearly subscription"
description: "Stop renewal for a Bearly subscription purchased on the web or through Apple's App Store, and confirm the cancellation date."
canonical: "https://bearly.ai/docs/account/cancel-subscription/"
markdown: "https://bearly.ai/docs/account/cancel-subscription.md"
---

# Cancel your Bearly subscription

HTML: [https://bearly.ai/docs/account/cancel-subscription/](https://bearly.ai/docs/account/cancel-subscription/)

> Stop renewal for a Bearly subscription purchased on the web or through Apple's App Store, and confirm the cancellation date.

Cancel the subscription through the payment source that sold it. Your Billing page identifies whether Bearly, Apple, or a team manages the plan.

## Subscription purchased on the web



1. Open the account menu and select **Billing & Pro**. You can also open **App Settings** → **Billing**.
2. Under **Current Plan**, select **Cancel Subscription**.
3. Confirm the cancellation in the secure billing page that opens.
4. Return to Bearly and reopen **Billing**. The plan should show **Cancels on** followed by the end date.



The plan normally remains available through the displayed billing-period end. To undo a scheduled cancellation before that date, select **Reactivate** under **Current Plan**.

## Subscription purchased in the iOS app



1. Open Bearly on the iPhone or iPad used for the subscription.
2. Open **App Settings** → **Billing**.
3. Under **Current Plan**, select **Cancel Subscription** or **Manage Account**.
4. In Apple's subscription screen, choose Bearly and confirm the cancellation.



If you are outside the iOS app, Bearly shows **Manage in iOS App** because Apple controls the subscription. You can also open the iPhone or iPad **Settings** app, select your Apple Account, then **Subscriptions** → **Bearly**.

> **Caution — Deleting Bearly does not cancel an App Store subscription**
> Cancel with Apple to stop future renewals. Account deletion can cancel a web-purchased Bearly subscription, but it cannot change an Apple subscription.


## Team-managed plan

If Billing shows **Team-Managed Billing**, the subscription belongs to the team. Ask a team administrator to change the plan, or select **Contact Support** for billing help. Removing your member account does not cancel the team's subscription.

## If the cancellation does not appear

- Reopen **Billing** and confirm you signed in to the Bearly account that purchased the plan.
- For an App Store purchase, select **Restore** in Bearly, then check Apple's subscription screen again.
- For a web purchase, select **Manage Account** to review the billing portal directly.
- If a renewal or charge still looks wrong, email [support@bearly.ai](mailto:support@bearly.ai) with the Bearly account email and the charge date. Do not email card details.

See [Manage billing and review usage](https://bearly.ai/docs/account/billing-usage.md) for plan changes, invoices, credits, and usage.

---

---
title: "Delete your Bearly account"
description: "Permanently delete your Bearly account, private work, credentials, and local Bearly data after reviewing subscriptions and retained billing records."
canonical: "https://bearly.ai/docs/account/delete-account/"
markdown: "https://bearly.ai/docs/account/delete-account.md"
---

# Delete your Bearly account

HTML: [https://bearly.ai/docs/account/delete-account/](https://bearly.ai/docs/account/delete-account/)

> Permanently delete your Bearly account, private work, credentials, and local Bearly data after reviewing subscriptions and retained billing records.

Account deletion is available in Bearly on web, desktop, iOS, and Android. Bearly checks subscriptions and shared work first, then verifies the request with a short-lived code sent to your account email.

> **Danger — Account deletion cannot be undone**
> Your chats, projects, documents, generated files, settings, saved prompts, phone profile, team membership, API keys, extension credentials, and Bearly sessions are permanently removed. Download anything you need before continuing.


## Delete the account



1. Open **App Settings** → **Account**.
2. Find **Delete account** and select **Delete**.
3. Review the lists under **Deleted** and **Kept as restricted records**. Also review any subscription warning or ownership blocker.
4. Select **Email verification code**.
5. Enter the six-digit code sent to the Bearly account email within 10 minutes, then type the complete email in **Account email**.
6. Select **Schedule account deletion**.
7. Save the deletion receipt and scheduled time shown on screen, then select **Done**.



Bearly emails the schedule to your account address and waits 24 hours before deleting the account. You can keep using Bearly during that time. An **Account marked for deletion** bar stays above the app and lets you select **Cancel deletion** until the final reminder. You can also cancel from **App Settings** → **Account** → **Delete**.

About one hour before deletion, Bearly sends a final email and closes cancellation. If an ownership blocker appears before deletion, Bearly leaves the account available and emails the action you need to take.

After the server completes the deletion, open Bearly clients sign out when they next contact the service and clear their local Bearly data. On desktop, this includes the isolated Bearly Browser profile. Existing Bearly sessions, API keys, and browser-extension credentials stop working. The deleted email address cannot be used to create another Bearly account. Bearly emails the deletion receipt after completion.

## Resolve a deletion blocker

Bearly will not delete shared work that could belong to someone else.

- If you administer a team, transfer administration or close the team first.
- If work you own has collaborators, transfer ownership or remove the collaborators first.
- If the blocker remains after making those changes, contact [support@bearly.ai](mailto:support@bearly.ai) from the account email.

## Check the subscription

- A web-purchased Bearly subscription is canceled when deletion runs, not when it is scheduled. If the subscription may renew during the 24-hour waiting period, [cancel it separately](https://bearly.ai/docs/account/cancel-subscription.md). Deletion does not refund unused time or credits.
- An App Store subscription is controlled by Apple and is not canceled by deleting Bearly. Use **Manage App Store subscription** in the deletion review, or [cancel the subscription separately](https://bearly.ai/docs/account/cancel-subscription.md), to stop future renewals.
- A team-managed plan belongs to the team. Deleting a member account removes its membership but does not cancel the team's subscription.

## Records Bearly keeps

Bearly keeps a limited set of billing, payment, subscription, usage-total, dispute, fraud-prevention, consent, tax, and legal records when they are needed for reconciliation or legal obligations. Bearly replaces the live account email in its retained operational ledgers with a restricted deletion reference. Payment processors may separately retain invoices and transaction records under their own legal obligations.

These retained records cannot be used to sign in or recover the deleted account. Bearly reviews them under its retention schedule and preserves them longer only when a legal or dispute hold requires it. The [Privacy Policy](https://bearly.ai/privacy-policy.md) governs retention and privacy rights.

If you cannot sign in, email [support@bearly.ai](mailto:support@bearly.ai) from the account address and ask for account deletion help.

---

---
title: "Privacy and security in Bearly"
description: "Understand how private AI works in Bearly, what a model receives, and which privacy, memory, encryption, and access controls you can use."
canonical: "https://bearly.ai/docs/privacy-security/"
markdown: "https://bearly.ai/docs/privacy-security.md"
---

# Privacy and security in Bearly

HTML: [https://bearly.ai/docs/privacy-security/](https://bearly.ai/docs/privacy-security/)

> Understand how private AI works in Bearly, what a model receives, and which privacy, memory, encryption, and access controls you can use.

Bearly gives you controls for deciding what context enters a private AI chat, protecting sensitive projects, and removing local or account data. This guide explains those visible controls. For a concise overview, read [Private AI in Bearly](https://bearly.ai/private-ai). For the legal terms governing personal information, read the [Privacy Policy](https://bearly.ai/privacy-policy.md).

## Decide what to share with a chat

Your message, the relevant conversation context, and anything you attach can be sent to the AI model handling your request. Enabled tools may bring in more information, such as local files, connected-account records, website content, or screenshots of your display. Running a tool locally does not mean its results stay only on your device.

Before sending sensitive work:

- Check the selected model and the files attached to the message.
- Remove information the task does not require.
- Check whether the chat is using **Standard** or **Full access**, and read any permission request before selecting **Allow**.
- Tell Bearly to pause before consequential actions such as sending, submitting, purchasing, or overwriting a file.

Full access skips ordinary tool-approval prompts for enabled external tools in this chat on this device. It does not connect accounts, enable disabled tools, or change the purposes for which Bearly uses personal information. See [Choose how Bearly uses external tools](https://bearly.ai/docs/settings/agent-permissions.md) to return to Standard or remove saved approvals.

Managed-team policies can limit the models and features available to a member. Organization administrators also control eligibility for Full access; granting eligibility does not activate a chat or give an administrator access to its private content. If a control is missing or locked, ask an organization administrator whether a policy applies.

## Limit chats to zero-retention models

Every model in the picker shows whether its provider retains your prompts and responses. Turn on **App Settings** → **Settings** → **Privacy** → **ZDR models only** to restrict chats, and their automatic fallbacks, to models whose providers keep nothing after processing a request.

See [Use zero-data-retention models](https://bearly.ai/docs/zero-data-retention.md) for what the setting covers and what it does not.

## Turn on unitemized billing

Turn on **App Settings** → **Settings** → **Privacy** → **Unitemized billing** to keep one rounded usage total per day instead of a record of each request's model, cost, and time. See [Turn on unitemized billing](https://bearly.ai/docs/unitemized-billing.md) for how it works and what it doesn't cover.

## Control personalization and memory

Open **App Settings** → **Chats** → **Personalization** to decide whether Bearly includes your profile in conversations. When **Personalization** is on, the **Name**, **Occupation**, and **About you** information you provide is shared with AI models in your conversations. Turn it off when you do not want that profile context included.

The **Memories** control determines whether a chat can draw on earlier Bearly conversations. In a project, you can limit memory to that project or allow searches across all chats. Select **Off** when the current conversation should not use remembered context.

Learn more in [Memory & Search](https://bearly.ai/docs/features/memory-search.md).

## Password-protect a project

Project password protection is useful when access to a project should require a separate password.

Open the project, select **Settings**, then select **Protect** beside **Encryption**. The password must contain at least eight characters.

> **Danger — Bearly cannot recover a forgotten project password**
> Without the password, the protected project and its data cannot be recovered. Store it somewhere appropriate before enabling protection.


You can select **Lock**, **Change**, or **Remove** beside **Encryption** from the same settings page. See [Project Encryption](https://bearly.ai/docs/projects/encryption.md) for the complete procedure.

## Review desktop access

In Standard, desktop features ask for approval when the requested access is not covered by a saved approval. Full access skips eligible Bearly action approvals.

- **Computer Use** can observe or control the desktop. Saved approvals have the duration you selected. **Stop** ends the current task without changing the permission mode or removing saved approvals; **Clear all permissions** returns to Standard and removes them.
- macOS may separately require **Accessibility**, **Screen Recording**, or microphone permission. Those operating-system permissions remain under your control in **System Settings**.
- A managed-team policy may disable desktop features even when the operating system permits them.

Close or hide unrelated sensitive windows before allowing full-desktop access. See [Use your computer with Bearly](https://bearly.ai/docs/features/computer-use.md) for its permission and stopping controls.

## Lock local access or sign out

Open your account menu and follow **App Settings** → **Account**.

- **Lock all data** locks password-protected work on the current device. You will need to enter the relevant passwords again.
- **Log out** clears local Bearly data and signs you out. You will need to sign in again.

Both actions show a confirmation before continuing.

## Delete your account or request privacy help

On web, desktop, iOS, and Android, open **App Settings** → **Account** and select **Delete** beside **Delete account**. Bearly shows what will be deleted and which restricted billing or compliance records must be retained, then sends a short-lived verification code to the account email. Verified deletion is scheduled 24 hours later and can be canceled in the same screen until Bearly sends the final one-hour reminder. See [Delete your Bearly account](https://bearly.ai/docs/account/delete-account.md) before continuing.

For a copy or correction of your personal information, deletion help when you cannot sign in, or another privacy question, contact [support@bearly.ai](mailto:support@bearly.ai).

## Read the governing terms

- [Privacy Policy](https://bearly.ai/privacy-policy.md) explains what information Bearly collects, why it is used, when it may be shared, retention, and privacy rights.
- [Terms of Service](https://bearly.ai/terms-of-service.md) explains the conditions for using Bearly.

If this guide and a legal policy appear to differ, the legal policy governs.

---

---
title: "Use zero-data-retention models"
description: "See which models run with zero data retention, restrict Bearly to those models, and understand what the setting covers."
canonical: "https://bearly.ai/docs/zero-data-retention/"
markdown: "https://bearly.ai/docs/zero-data-retention.md"
---

# Use zero-data-retention models

HTML: [https://bearly.ai/docs/zero-data-retention/](https://bearly.ai/docs/zero-data-retention/)

> See which models run with zero data retention, restrict Bearly to those models, and understand what the setting covers.

A zero-data-retention (ZDR) route means the provider handling your request does not keep your prompt or the model's response after it finishes processing them. Bearly labels every model in the picker and gives you one switch that limits chats to ZDR models only.

## Check a model's retention status

Open the model picker in a chat. Each model carries a badge beside its name:

- **ZDR** — the provider on this model's route does not store your prompts or responses after processing.
- **Standard** — this model's route has no verified zero-data-retention guarantee.

Hover the badge, or hold it on a touch screen, to read what the label means for that model.

A model is labeled **ZDR** only when Bearly has verified the retention terms for the route it actually uses and tested that route in production. Anything unverified is labeled **Standard**, so an unknown answer never appears as a privacy guarantee.

## Restrict Bearly to ZDR models

Open your account menu and follow **App Settings** → **Settings** → **Privacy**, then turn on **ZDR models only**.

While the setting is on:

- The model picker lists only models with a verified zero-data-retention route. Models without one are not offered.
- Automatic fallbacks follow the same rule. When Bearly retries your message on a different model because the first one is unavailable, it uses another ZDR route.
- A conversation still set to a model without a ZDR route runs on a comparable model that has one, rather than sending your message to the original.

The setting belongs to your account rather than one device, so it applies in Bearly chat on web, desktop, iOS, and Android once they sync.

> **Note — The setting fails closed**
> If Bearly cannot reach your synced settings, it enforces ZDR-only routes until the connection recovers, and the switch is unavailable until then. A sync problem never quietly downgrades your privacy choice.


## What the setting covers

**ZDR models only** governs the models that write chat replies, including the reasoning and tool-calling steps Bearly runs inside a conversation.

It does not change these, which run on separate models and services:

- Image generation and editing, video generation, text to speech, transcription, and text extraction from uploaded documents.
- Tools that reach outside services on your behalf, such as web search, maps, and connectors. Those send the query they need to the service handling it.

When a task should not reach outside services at all, set **Web Search** to **Off** in [chat settings](https://bearly.ai/docs/settings/chat-settings.md) and keep each service **Off** under **Connectors** for that chat. Tools such as Maps and Weather have no per-chat switch; a team administrator can turn them off with a [team policy](https://bearly.ai/docs/teams/policies.md).

## Teams

**ZDR models only** is a personal setting; each member turns it on for their own account.

A team administrator can separately decide which models members may use at all. Under **Models** in a team policy, each model row carries the same **ZDR** badge, so an administrator can build an allow-list of zero-data-retention models for the whole team. See [Team Policies](https://bearly.ai/docs/teams/policies.md).

## Related

- [Privacy and security in Bearly](https://bearly.ai/docs/privacy-security.md) — the full set of privacy controls, including personalization, memory, and project passwords.
- [Privacy Policy](https://bearly.ai/privacy-policy.md) — the terms governing personal information, sharing, and retention.

---

---
title: "Turn on unitemized billing"
description: "Turn on unitemized billing so your usage history keeps one rounded total per day instead of a line for every request."
canonical: "https://bearly.ai/docs/unitemized-billing/"
markdown: "https://bearly.ai/docs/unitemized-billing.md"
---

# Turn on unitemized billing

HTML: [https://bearly.ai/docs/unitemized-billing/](https://bearly.ai/docs/unitemized-billing/)

> Turn on unitemized billing so your usage history keeps one rounded total per day instead of a line for every request.

With **Unitemized billing** on, Bearly bills your usage as one rounded total per day instead of itemizing each request's model, cost, and time.

The setting is available on paid plans. On a managed team, an administrator has to allow it before members can turn it on.

## Why turn it on

A per-request usage history shows which AI models you use, when you use them, and what each request cost. That detail helps answer billing questions, but it is also a record of how you work. With unitemized billing, your plan's limits and billing use rounded daily totals, and the per-request usage detail isn't kept in your usage history.

## Turn on unitemized billing



1. Open your account menu and follow **App Settings** → **Settings** → **Privacy**.
2. Turn on **Unitemized billing**.
3. Read the notes, check both boxes, and select **Turn on**.



New usage then appears in **App Settings** → **Settings** → **Usage** as one **Bearly Private** line, with no model breakdown or call count.

To stop, turn the switch off. New usage is recorded per request again, and the daily totals already saved stay as they are.

## How it works

- **One running total per day.** Each request adds its cost to that day's total. The request's model, size, and time aren't stored in your usage history.
- **Only the final total is saved.** During the day, the running total is kept in a part of our database that isn't copied into backups. When the day ends, Bearly saves the final total, rounded to the nearest 10 cents, or to the cent when it's under 10 cents.
- **Limits work the same way.** Your daily and monthly limits count these totals. A few kinds of usage with their own limits, such as Deep Research, keep separate daily totals.
- **No per-request analytics.** The Bearly app and servers stop sending per-request usage analytics for your account.
- **Provider bills are checked without your account.** Bearly still pays each AI provider for the requests it handles. To check those bills, it keeps per-model totals that combine everyone using unitemized billing, with no account attached. A day's per-model totals are saved only when at least 10 accounts used unitemized billing that day; otherwise they're added to totals for the whole month.

## What stays the same

- **Your chats.** Conversations are saved as usual, including which model answered each message and when. Delete any you don't want kept.
- **What the AI provider receives.** The provider that answers a request still receives what it needs to answer. To use only providers that keep nothing after processing, turn on [zero-data-retention models](https://bearly.ai/docs/zero-data-retention.md).
- **Earlier usage.** Usage from before you turned the setting on keeps its detail, including in backups until they expire.

## Tradeoffs

- **Billing is final.** Bearly can't review or refund individual charges, because it doesn't have them.
- **Totals are rounded.** Each saved total can differ from its exact cost by up to 5 cents.
- **A quiet day says more.** When a day has a single request, or one large request such as a Deep Research run, its total still shows roughly what that request cost.
- **Combined totals aren't anonymous in every case.** If only one account used a particular model on a day, or few accounts use unitemized billing at all, the combined per-model totals say more about those accounts.
- **Older app versions.** Older versions of Bearly keep sending usage analytics until they update.
- **Free plan.** On the Free plan the setting is paused and usage is recorded per request.
- **Web server logs.** Bearly's web servers keep access logs of incoming requests, including IP address and time, for security.

## For team administrators

Open your account menu, select **Organizations**, choose your organization, and open **Policies**. Select a policy or **Add policy**. Under **Feature access** → **Privacy & Security**, turn on **Unitemized billing**, then select **Save**. See [Set organization policies](https://bearly.ai/docs/teams/policies.md).

- Each member still decides whether to turn it on for their own account.
- If you stop allowing it, members' unitemized billing pauses and new usage is recorded per request again.
- Team usage reports show a member's daily totals as **Bearly Private**, without per-model detail or call counts. See [Review organization usage](https://bearly.ai/docs/teams/usage.md).

## Related

- [Use zero-data-retention models](https://bearly.ai/docs/zero-data-retention.md)
- [Privacy and security in Bearly](https://bearly.ai/docs/privacy-security.md)
- [Manage billing and review usage](https://bearly.ai/docs/account/billing-usage.md)

---

---
title: "Changelog"
description: "Every Bearly release since June 2024, newest first, with what changed in each one."
canonical: "https://bearly.ai/docs/changelog/"
markdown: "https://bearly.ai/docs/changelog.md"
---

# Changelog

HTML: [https://bearly.ai/docs/changelog/](https://bearly.ai/docs/changelog/)

> Every Bearly release since June 2024, newest first, with what changed in each one.

Every Bearly release since June 2024, newest first. These notes also appear in the app under **What's New**.

## 4.47 — September 27, 2026

**Highlights**

- **📺 Watch and check YouTube videos in chat** — Paste a link to get the video with a verdict on whether to watch it, key moments, shorter cuts, and a summary you can jump through by the second. [Blog](/blog/check-a-youtube-talk-before-you-watch-it) · [Docs](https://bearly.ai/docs/features/youtube.md)

**Also new**

- **📚 Mentioned** — See the papers, books, repositories, and tools a video names, with the moment each comes up, and look any of them up.
- **🔢 Check claims** — Check the figures a video states and see how many sites give the same figure or a different one.
- **📋 Copy with timestamps** — Copy a video's card as text, with each quote linked to the second it was said.

**Fixes & improvements**

- Get Video follows your organization's File uploads setting, since it saves a file to the chat.

## 4.46 — September 25, 2026

**Highlights**

- **🛡️ Full access for agents** — Choose Full access for one chat on one device to let enabled tools act without asking for approval each time. [Docs](https://bearly.ai/docs/settings/agent-permissions.md)

**Also new**

- **⏰ Full access for Routines** — Separately approve a routine's selected tools to work while you're away, with a new review when its instructions or tools change.
- **🏢 Agent access for teams** — Organization admins decide who can enable Full access through a team default and individual member exceptions.

**Fixes & improvements**

- Full access in chats pauses after an app restart or 24 hours away, with clear controls to resume or return to Standard.
- Shared Pages can finish preparing even when artwork generation fails.
- Gemini image requests no longer use your allowance when no image is returned.
- Saved Kimi K2.6 selections now use GLM 5.3.
- Clicking a Page preview now closes its publishing menu.

## 4.45 — September 22, 2026

**Highlights**

- **⏰ Routines can use a browser** — Turn on Browser access to let a routine use Bearly Browser or its own tabs in your Chrome while you're away, including submitting forms, after a one-time warning. [Docs](https://bearly.ai/docs/features/routines.md)

**Also new**

- **🛡️ Browser in Routines** — Team admins can allow or block browser use in Routines separately from Browser in chats.

**Fixes & improvements**

- Turning on a routine's unattended local or browser access now always takes the on-screen card; a spoken yes can't approve it.

## 4.44 — September 22, 2026

**Highlights**

- **New! GPT 6 Sol, GPT 6 Luna, Opus 5.5, and more** — Try GPT 6 Sol, GPT 6 Luna, Claude Opus 5.5, Grok 4.7, GLM 5.3 and Flash, plus Bearly Fast and Ultra—with GPT 6 Sol and Opus 5.5 at half or less than GPT 5.6 Sol and Fable 5.1 per token, respectively.

**Also new**

- **🗂️ Archives view** — See archived chats and projects in one date-grouped list, filter larger archives, and unarchive several at once.
- **🎨 Project icons** — Choose an emoji for each project or use Bearly's suggestion when you create one.
- **🖼️ New image models** — Create and edit with MAI Image 2.6, Qwen Image 3 Pro, Grok Imagine 2.0, and Muse Image 1.0.
- **🧠 System One Jev API** — Use Jev Latest, Preview, or 1.13 through the Bearly API with zero data retention.

**Fixes & improvements**

- Get Video now handles public Vimeo links, including longer videos that need a lower-resolution download.
- Pinned chats, archives, read state, and other synced settings save more reliably if you reload or close the page right after a change.
- The image editor appears with the latest result, and earlier image results can jump to it.
- Image generation and editing follow refined prompts more closely, including area edits, edge extensions, and image combinations.
- HEIC and HEIF photos upload to the image gallery more reliably in Chrome.
- Deep Research shows clearer errors when a run cannot start or continue.
- Desktop sign-in offers a retry when the app does not confirm that it saved your session.
- New accounts follow the device's appearance setting by default.

## 4.43 — September 15, 2026

**Highlights**

- **🔗 Share a chat or file as a Page** — Turn a conversation or file into a polished Page at a stable link, with access for anyone with the link or selected people and teams. [Blog](/blog/share-the-work-from-your-chat) · [Docs](https://bearly.ai/docs/chats/share.md)
- **🎨 Create and edit images in one place** — Create, edit, change an area, extend edges, and combine images from one action bar in the chat, with several runs at once.

**Also new**

- **📞 Choose your voice model** — Pick GPT-Live 1, Gemini 3.8 Live, or Gemini 3.8 Live Extended Thinking under Settings → Chats → Voice on paid plans; the call shows which model it runs on.
- **🗣️ Spoken approvals** — Answer permission questions out loud during a voice call, including don't ask me again for the chat.
- **🏢 Access banner** — Team admins can require members to acknowledge a message each session, day, or week, with an optional sign-out action.
- **💳 Keep chatting on Bearly Pro** — When your plan allowance and credits run out, chat continues on Bearly Pro until credits are restored.

**Fixes & improvements**

- Copy link is the primary share action, and Page saving is faster.
- Browser and Computer Use actions confirm consequential steps and recover more reliably.
- Sign-in and public pages load faster.
- Get Desktop App links point to the current release.

## 4.42 — September 10, 2026

**Highlights**

- **📚 Search academic papers with Scholar** — Find papers, read available passages, and follow citations and related research. [Blog](/blog/ask-what-the-research-says) · [Docs](https://bearly.ai/docs/features/scholar.md)
- **🌐 Publish a Page** — Share an interactive Page at a stable bearly.sh link, with access for anyone with the link or selected people and teams.
- **📞 Voice calls run on GPT-Live** — Bearly listens while it speaks, so you can cut in, and a call keeps going while Bearly uses tools in the chat. [Blog](/blog/voice-calls-now-run-on-gpt-live) · [Docs](https://bearly.ai/docs/features/voice-conversations.md)

**Also new**

- **👥 Group admins** — Delegate group management, review group usage, and find shared projects.
- **✅ Approval duration** — Choose approvals for one action, a response, a session, or the saved chat, and revoke them when needed.
- **⬇️ Browser downloads** — Bring approved downloads from the desktop Browser into the conversation as files Bearly can use.

**Fixes & improvements**

- GPT-6 Astra and Gemini 3.8 Flash are available.
- GPT Image 2.5 Flare and Sunburst support image generation and editing.
- Project files can be renamed and downloaded directly from Docs & Data.
- Desktop Browser tabs stay active when switching chats or closing the panel.
- Extension page summaries recover when screenshot capture is unavailable.
- Chat text and math stay within the conversation on narrow screens.
- Gemini handles images, audio, and video from links more reliably.

## 4.41 — September 2, 2026

**Highlights**

- **🎙️ Dictate instead of typing** — Hold a key or use the message-box control to turn speech into editable text, with optional cleanup. [Blog](/blog/hold-a-key-and-talk) · [Docs](https://bearly.ai/docs/features/dictation.md)
- **🌐 Bearly works in your own Chrome** — Use the extension beside a page, or let a desktop chat work in Chrome tabs where you're already signed in. [Blog](/blog/bearly-works-in-your-own-chrome) · [Docs](https://bearly.ai/docs/features/chrome-control.md)

**Also new**

- **📐 Visualize** — Turn an idea, layout, process, or dataset into a focused visual, interactive planner, or simulation.
- **📋 Lead enrichment** — Enrich up to ten known leads at once from profile links, email addresses, names and employers, or roles and companies.
- **🏢 Microsoft Entra provisioning** — Managed-team admins can sync assigned users and groups from Microsoft Entra.

**Fixes & improvements**

- Claude Fable 5.1 is available.
- Deep Research now uses actual cost against your allowance, with a per-run cap and one run at a time.
- Background refreshes now wait when answers, unsent drafts, pending files, or meeting recordings are active.
- New Chat and standard Browser shortcuts work reliably, and deleted draft chats recover into a fresh chat.

## 4.40 — August 26, 2026

**Highlights**

- **🔎 Find a name, then check the domain** — Ask Bearly for domain ideas, then check whether an exact domain is available to register. [Blog](/blog/find-the-name-then-check-the-domain) · [Docs](https://bearly.ai/docs/features/domain-research.md)
- **💳 Team billing and seats** — Buy and manage Pro, Analyst, Research, or Max seats from one team account. [Blog](/blog/give-each-teammate-the-right-bearly-plan) · [Docs](https://bearly.ai/docs/teams/billing.md)

**Also new**

- **⏰ Models for Routines** — Choose which model runs each schedule.
- **🏢 Automatic team joining** — Add anyone with your managed email domain automatically, or require an administrator to add each person.

**Fixes & improvements**

- Routines create a new chat and keep running if their original chat is deleted.
- Longer multi-step work can keep going, with a compact live status and expandable details.
- Chat lists refresh more reliably across devices.
- Project sharing finds individual teammates more reliably.
- Custom prompts, input references, and desktop notifications are more reliable.

## 4.39 — August 19, 2026

**Highlights**

- **🛡️ Limit chats to zero-retention models** — Turn on ZDR models only in Privacy settings to use just the models whose providers keep nothing after answering. [Blog](/blog/keeping-your-work-private-zero-data-retention) · [Docs](https://bearly.ai/docs/zero-data-retention.md)

**Also new**

- **🏷️ Retention badges** — Every model in the picker shows whether its provider keeps your prompts and responses.

## 4.38 — August 17, 2026

**Also new**

- **📊 Charts for every account** — Turn prepared spreadsheet data into an interactive KPI, bar, or line chart.
- **📱 Phone & SMS** — Add a US or Canadian number, then ask Bearly to text you from chats and Routines.

## 4.37 — August 16, 2026

**Highlights**

- **📊 Chart a spreadsheet in Bearly** — Attach a spreadsheet and get an interactive KPI, bar, or line chart with the source data ready to inspect and download. [Blog](/blog/chart-a-spreadsheet-in-bearly) · [Docs](https://bearly.ai/docs/features/charts.md)

**Also new**

- **🧭 Chat timeline** — Jump between prompts and chapters in longer conversations.
- **🔎 Search chats and files** — Find conversations by message text or the names of uploaded and generated files, with dates for matched messages.
- **🌐 Better Browser control** — Bearly can hover, drag, scroll nested pages, and use screenshots to work with harder page layouts.
- **👥 Team directories** — Managed-team admins can choose whether members can find coworkers by name or work email.

**Fixes & improvements**

- Gemini 3.7 Flash and Grok 4.6 are available.
- Grok text, image, and voice requests now run with zero data retention.
- Charts, research, images, and other visible results remain available as Bearly continues multi-step work.
- Long conversations preserve active instructions and important context more reliably as they grow.
- Large team roster imports are faster and flag duplicate or wrong-domain rows more clearly.
- Tool failures explain more clearly what to retry or change.
- Notifications are more consistent and take up less space.
- Math stays within the chat on narrow screens, and text fields no longer trigger unwanted zoom on iPhone or iPad.
- Deep Research email delivery, desktop Skills, and Windows Computer Use setup are more reliable.
- New Chat, project folder sync, and expanded Browser controls recover more reliably.

## 4.36 — August 10, 2026

**Also new**

- **📊 Editable tables** — Edit CSV, TSV, and other delimited files in a full-size grid, with every save kept as a version.

**Fixes & improvements**

- Generated file cards and full previews now share the same version history, including newer versions created later.
- Every inline file preview has a download control, with richer previews for more file types.
- Long chats stay steady instead of flashing or jumping when you send a message.
- Local meeting recordings finish transcribing more reliably.
- Chat titles appear sooner and recover more reliably when generation fails.
- The desktop Browser and chat sidebar no longer compete for space.
- Document uploads and model responses recover from more temporary failures.
- People Search no longer shows broken text when it doesn't find published work.
- Login codes, scheduled results, and other Bearly emails have a more consistent design.

## 4.35 — August 6, 2026

**Also new**

- **📬 Message queueing** — Send follow-ups while Bearly is answering; they wait in the composer and send one at a time.

**Fixes & improvements**

- ⌘Enter sends a message immediately, interrupting the current answer.
- ↑ in an empty composer recalls your most recent queued message.

## 4.34 — August 5, 2026

**Highlights**

- **🔌 Connect your own MCP servers** — Add an HTTPS MCP server on the web or desktop, or run an npm MCP package in the desktop app, and use its tools in chats and Routines. [Docs](https://bearly.ai/docs/features/connectors.md)

**Fixes & improvements**

- Team policies can provide MCP servers to members and decide whether members can add their own.
- Connectors now show why a server failed, and a local server that stopped restarts the next time a chat needs it.
- Servers and keys you add stay in the browser or desktop profile where you set them up.

## 4.33 — August 4, 2026

**Highlights**

- **🔎 People Search** — Meeting someone new? Bearly makes sure it has the right person, then writes up their public work with a source and a date on every claim. [Blog](/blog/find-the-person-then-the-footprint) · [Docs](https://bearly.ai/docs/features/people-search.md)
- **🌐 Bearly Browser** — Open a browser you and Bearly share in the desktop app, and let it read and act on pages with you. [Blog](/blog/a-browser-beside-the-conversation) · [Docs](https://bearly.ai/docs/features/browser.md)

**Also new**

- **🖼️ Image search** — Ask for images and get a gallery you can open, copy, or download.
- **📨 Email & text from chat** — Ask Bearly to email or text you results from any conversation.
- **🔌 Connectors on desktop** — Use your connected apps in the desktop app, not just on the web.
- **📁 Meeting transcripts in Projects** — Send a meeting transcript to a project so chats there can use it.

**Fixes & improvements**

- Claude Opus 5 is available.
- Voice conversations, transcription, and text-to-speech use upgraded audio models.
- Web search cites sources as compact numbered links.
- Chats sync more safely through connection problems, and long chats render faster.
- The desktop app shows a clearer screen when it can't connect and recovers on its own.
- Scanned documents, connector sign-ins, sidebar menus, and iOS input are more reliable.

## 4.32 — July 26, 2026

**Highlights**

- **🎙️ Record meetings with live transcripts** — Record a meeting, call, or voice note and watch the speaker-aware transcript arrive live. [Blog](/blog/record-meetings-in-bearly) · [Docs](https://bearly.ai/docs/features/meetings.md)

**Also new**

- **🗺️ Maps** — Ask for places, reviews, or directions by car, foot, bike, or transit.
- **🌤️ Weather** — Ask for forecasts, severe-weather alerts, air quality, or pollen anywhere.
- **💬 Chat management** — Find, read, rename, and organize chats with confirmation for sensitive actions.

**Fixes & improvements**

- Math renders without stray dollar signs.
- Routines has a clearer schedule builder and week view.
- The new-chat box opens at its full height.

## 4.31 — July 22, 2026

**Also new**

- **⏰ Routines** — Deliver scheduled prompts to their own chat, email, or text.
- **🔎 SEO Data** — Ask for current demand, search results, rankings, competitors, backlinks, or page audits.
- **🧭 Faster references** — Type `/` or `@` to add files, skills, prompts, research, and tools to a message.
- **🎥 Video understanding** — Inspect uploaded videos by scene, audio, and time range.
- **📸 Better X research** — Bring post images and attached videos into the analysis.
- **🔐 Password-protected PDFs** — Unlock protected files during upload.

**Fixes & improvements**

- New chats recover from connection failures without losing drafts or creating duplicates.
- Research sources, partial answers, file errors, iOS voice mode, desktop startup, and chat titles are more reliable.

## 4.30 — July 21, 2026

**Highlights**

- **⏰ Routines** — Put a prompt on a schedule and receive each result in chat, by email, or by text. [Blog](/blog/put-bearly-on-a-schedule) · [Docs](https://bearly.ai/docs/features/routines.md)

## 4.29 — July 3, 2026

**Also new**

- **📝 Transcripts as files** — Save audio and video transcripts as reusable, searchable text files in chat.

**Fixes & improvements**

- Re-transcribing the same recording reuses its existing file.

## 4.28 — June 30, 2026

**Highlights**

- **Bearly Pro, Claude Sonnet 5, and Nano Banana Lite** — New language and image options balance frontier quality, speed, and price.

**Also new**

- **✨ Needs Attention** — Gather unread chats and pending approvals in one sidebar filter.
- **📄 Better OCR** — Preserve tables and retry low-confidence scans automatically with Mistral OCR 4.
- **💻 Local approval controls** — Grant or revoke command auto-approval directly from chat.

**Fixes & improvements**

- Text files skip OCR for faster, more reliable uploads.
- Chat forks, iOS input, local folder identity, and new-chat tips are more reliable.

## 4.27 — June 17, 2026

**Also new**

- **🎙️ Better text-to-speech** — Choose a natural voice and steer its tone, pace, or accent.

## 4.26 — June 17, 2026

**Also new**

- **🗂️ Safer archives** — Filter clearly marked archived chats and restore one before deleting it.

**Fixes & improvements**

- The Archive control disappears correctly when no archived chats remain.

## 4.25 — June 9, 2026

**Highlights**

- **Claude Fable 5** — Anthropic's flagship model supports long context, images, tools, and adaptive reasoning.

## 4.24 — June 4, 2026

**Highlights**

- **🎬 Generate and edit video** — Create video from text or images, add native audio, or transform an existing clip by describing the change.

**Also new**

- **🎚️ Video controls** — Choose length, resolution, aspect ratio, and optional start or end frames.

**Fixes & improvements**

- Video replies no longer expose internal control tags.

## 4.23 — May 28, 2026

**Highlights**

- **Claude Opus 4.8** — Anthropic's latest flagship model replaces earlier Opus versions in Bearly.

## 4.22 — May 20, 2026

**Highlights**

- **🔬 Deep Research** — Review a research plan, then receive a thorough source-backed report.
- **Gemini 3.5 Flash** — Google's fast, affordable everyday model is available.

**Fixes & improvements**

- Apple-billed customers can buy credit boosts on the web.
- Copied answers preserve formatting in Word, Outlook, and other rich-text apps.
- Voice mode works better in noisy rooms.
- Desktop navigation, spelling suggestions, and app stability are improved.
- Projects replace legacy document collections for organized file workflows.

## 4.21 — May 13, 2026

**Highlights**

- **🎙️ Voice mode** — Talk with Bearly in real time, then continue from the transcript saved in chat.

## 4.20 — May 10, 2026

**Also new**

- **🖼️ Cleaner media workflows** — Review image galleries and inline media without leaving chat.
- **🎨 Themed Pages** — Generated Pages inherit your Bearly colors.
- **⬇️ Office downloads** — Save either a PDF preview or the original file.
- **🧩 Local skill scripts** — Run approved preset scripts without exposing a general shell.

**Fixes & improvements**

- Cowork approvals, local skills, desktop downloads, and runtime settings are clearer.
- Sidebar attention states, encrypted-project sync, mobile input, and desktop startup are more reliable.

## 4.19 — May 5, 2026

**Highlights**

- **💻 Cowork** — Let Bearly read, write, and run approved commands inside a folder you choose. [Docs](https://bearly.ai/docs/features/work-locally.md)
- **🧩 Skills** — Install reusable abilities from Git, GitHub, or Bearly's built-in collection.
- **GPT-5.5 and Grok 4.3** — New frontier reasoning models are available.

**Fixes & improvements**

- Tool calls and thinking steps use a compact, inspectable row.
- Tool arguments appear while Bearly is preparing them.
- Desktop menus, international text input, and document rendering are more reliable.

## 4.18 — April 29, 2026

**Highlights**

- **🏛️ Council of Experts** — Ask several leading models at once and receive one bias-reduced synthesis.
- **🐦 X research** — Search public X posts and profiles for current discussion, sentiment, and expert commentary.

**Also new**

- **🎯 Tidier agentic view** — Inspect thinking and tool steps from one compact row.

## 4.17 — April 21, 2026

**Highlights**

- **Kimi 2.6 and GPT Image 2** — New agentic and image-generation models are available.

## 4.16 — April 20, 2026

**Highlights**

- **🚀 A faster new-chat page** — Bearly opens to a dedicated home base that starts a conversation without a spinner.

**Also new**

- **🎞️ Grounded media answers** — Questions about audio and video start from a transcript.

**Fixes & improvements**

- Billing, syncing, and app stability are smoother.

## 4.15 — April 16, 2026

**Highlights**

- **Claude Opus 4.7** — Anthropic's latest flagship model is available.

**Also new**

- **⏭️ Interrupt & Send** — Send a new message to stop and redirect an answer in progress.
- **📸 Mobile camera** — Open the camera from chat with one tap.
- **🖼️ Upload previews** — See image thumbnails and clearer processing progress before sending.

**Fixes & improvements**

- PDFs, downloads, file actions, video, and Markdown math work more reliably on mobile.

## 4.14 — April 14, 2026

**Highlights**

- **🧠 Conversation history modes** — Choose dynamic trimming for chat or compaction for long, agentic work.

**Also new**

- **💅 Cleaner chat design** — Use flatter messages and smaller citation markers.
- **🔎 Smarter web fetch** — Search several queries at once and save large sources as attachments.
- **📄 Better document handling** — Read documents directly and turn large text pastes into files.
- **💰 Usage summary** — See spend, daily rate, boost balance, and credit reset date together.

**Fixes & improvements**

- Streaming, Markdown, code blocks, navigation, and long chats are more reliable.

## 4.13 — April 8, 2026

**Also new**

- **📌 Chat pinning** — Keep important conversations at the top of the sidebar.
- **🧠 Smarter conversation memory** — Preserve long chats through background compaction and better caching.
- **📸 Better image uploads** — Reliably upload large photos, HEIC files, and WebP images.

**Fixes & improvements**

- The sidebar sorts chats by recent use.
- OpenAI responses stream faster and cached follow-ups cost less.
- Chat titles, image galleries, metadata handling, and model menus are more reliable.

## 4.12 — April 5, 2026

**Also new**

- **🔎 Smarter web research** — Follow evidence beyond snippets and verify it through clearer sources.
- **✉️ Send email from chat** — Email a summary, notes, or attachments from the current conversation.
- **🖼️ Image understanding** — Inspect images found during a workflow and keep working with them.
- **🧠 Longer workflows** — Continue complex, multi-step tasks through more work before stopping.

## 4.11 — April 2, 2026

**Highlights**

- **🧠 Memory recall** — Bearly can bring relevant details from past conversations into the current answer.

**Also new**

- **📄 Deeper document understanding** — Search inside uploads and walk through them section by section.

## 4.10 — March 31, 2026

**Highlights**

- **🪪 Personalization** — Tell Bearly about your role and preferences so responses fit your work.

**Also new**

- **💰 Credit boosts** — Add one-time credits from Billing when you need more before reset.
- **✍️ Canvas preview** — Edit raw Markdown beside a live rendered preview.
- **📊 Model usage** — See exactly where your credits go by model.
- **🖼️ Better image viewer** — Zoom, pan, pinch, and quickly reset an image.

**Fixes & improvements**

- Answers are shorter and Markdown spacing is tighter by default.

## 4.8 — March 4, 2026

**Highlights**

- **🎨 Themes** — Choose from 11 themes that restyle chat, navigation, code, and every page.

**Also new**

- **✨ Fresh new look** — A consistent design system now runs across every page and component.
- **💬 Smoother chat** — Media keeps playing while answers stream and switching chats is flicker-free.

**Fixes & improvements**

- Chat scrolling is smoother.
- Markdown and code rendering are clearer and more accurate.

## 4.7.2 — February 26, 2026

**Highlights**

- **Gemini 3.1 Pro and Nano Banana 2** — Google's latest language and image models are available.

**Also new**

- **📁 Chat file browser** — Browse and download uploaded, generated, and project files from chat.
- **✍️ More response styles** — Use Classic or No Style and set project-level defaults.

**Fixes & improvements**

- Desktop folder picking and multi-monitor window placement are more reliable.

## 4.7.1 — February 17, 2026

**Highlights**

- **🎨 Response styles** — Choose how Bearly writes in each chat, from concise to explanatory or formal.
- **Claude Sonnet 4.6** — Anthropic's latest Sonnet model is the new featured model.

**Also new**

- **⚙️ Redesigned Settings** — Change this chat or future chat defaults from a faster modal.
- **🧭 Better navigation** — Use project breadcrumbs, global search, and one app menu for common controls.

**Fixes & improvements**

- Recommended automatically chooses the best available model for new chats.
- Images, desktop popovers, and the model picker render more reliably.

## 4.7 — January 6, 2026

**Highlights**

- **🏞️ Image editing gallery** — Generate, edit, and combine images in one focused workspace. [Docs](https://bearly.ai/docs/features/image-generation.md)
- **💅 Grammar corrections** — Review and apply sentence-level writing corrections inside Bearly. [Docs](https://bearly.ai/docs/features/grammar-correction.md)

**Fixes & improvements**

- Image generation adds SeedDream 4, Qwen, Imagine 4, and Flux 2.
- The app loads faster and syncs chats more reliably.

## 4.6.3 — November 24, 2025

**Highlights**

- **Claude Opus 4.5** — Anthropic's latest Opus model is available.

**Fixes & improvements**

- Gemini 3.1 Flash Image is the new default image-editing model.

## 4.6.2 — November 18, 2025

**Highlights**

- **Gemini 3** — Google's latest general-purpose model is available.

## 4.6.1 — November 14, 2025

**Highlights**

- **GPT-5.1 and GPT-5 Pro** — OpenAI's latest models are available in Bearly.

## 4.6.0 — October 23, 2025

**Also new**

- **🗂️ Prompt Templates** — Reusable prompts now have their own page.

**Fixes & improvements**

- The sidebar and several pages have a clearer, more streamlined design.
- Answers stream more smoothly.
- Cross-device sync is faster and more reliable.

## 4.5.2 — September 15, 2025

**Highlights**

- **🔐 Project encryption** — Protect an entire project with a password only you know. [Docs](https://bearly.ai/docs/projects/encryption.md)

**Fixes & improvements**

- Copied answers paste more reliably into Google Docs and other rich-text editors.
- The mobile sidebar and Artifact experience are more stable.

## 4.5.1 — September 3, 2025

**Highlights**

- **Gemini 2.5 Flash Image** — Google's Nano Banana model is available for image editing. [Docs](https://bearly.ai/docs/features/image-editing.md)

**Fixes & improvements**

- Audio, video, scrolling, media rendering, and cross-device sync are more reliable.

## 4.5 — August 25, 2025

**Highlights**

- **📚 Project documents** — Build a shared knowledge base that every chat in a project can use automatically. [Docs](https://bearly.ai/docs/projects/documents.md)

**Also new**

- **📄 Richer document processing** — Preserve more information from images, tables, charts, and page layout.

**Fixes & improvements**

- Word, Excel, PowerPoint, and other office files process faster and more reliably.
- GPT-5 formats code more reliably.

## 4.4.1 — August 15, 2025

**Highlights**

- **GPT-5** — OpenAI's adaptive reasoning model is the new Bearly default.

**Fixes & improvements**

- Claude Opus was upgraded to 4.1.
- Chat scrolling and performance are smoother.

## 4.4 — August 1, 2025

**Highlights**

- **📁 Projects** — Group related chats and give each project its own settings and workflow. [Docs](https://bearly.ai/docs/projects/overview.md)

## 4.3.2 — July 20, 2025

**Highlights**

- **Grok 4** — xAI's reasoning model can work with images, code, and complex questions.

## 4.3.1 — July 10, 2025

**Highlights**

- **🎥 Video generation** — Turn an image and prompt into a five-second video with Kling AI 2.1. [Docs](https://bearly.ai/docs/features/video-generation.md)

**Also new**

- **🎨 More image editing** — Compare Flux Kontext and GPT Image results.

**Fixes & improvements**

- OpenAI GPT-4.5 was retired.

## 4.3 — June 25, 2025

**Highlights**

- **🧠 Memory and chat search** — Search encrypted conversations and let Bearly recall useful context from them. [Docs](https://bearly.ai/docs/features/memory-search.md)

**Also new**

- **⬇️ Artifact downloads** — Export an Artifact to run or deploy outside Bearly.
- **🧭 Chat navigation** — Move through chats from a dedicated top bar.

**Fixes & improvements**

- Chat scrolling is more reliable.
- Model pricing reflects recent provider price reductions.
- Canvas and Artifact windows behave more reliably.

## 4.2.1 — June 15, 2025

**Highlights**

- **Claude 4 models** — Claude Sonnet 4, Sonnet 4 Reasoning, and Opus 4 are available.

## 4.2 — June 1, 2025

**Highlights**

- **🎨 Edit, combine, and inpaint images** — Restyle an image, merge several images, or paint over the exact area you want to change. [Docs](https://bearly.ai/docs/features/image-editing.md)

**Also new**

- **🗑️ Bulk delete** — Select and remove several chats at once.

**Fixes & improvements**

- Gemini 2.5 Flash is available.

## 4.1.1 — April 17, 2025

**Highlights**

- **🧱 Artifacts** — Create web apps, reports, games, and other interactive work directly in chat. [Docs](https://bearly.ai/docs/features/artifacts.md)
- **🐍 Code Interpreter** — Run Python for analysis, calculations, and file or image transformations. [Docs](https://bearly.ai/docs/features/code-interpreter.md)

**Also new**

- **🔊 Read Aloud** — Listen to any message directly from chat.

**Fixes & improvements**

- Images expand when clicked.
- Chats preserve drafts, running queries, and scroll position more reliably when you switch away.
- Gemini 2.5 Pro, OpenAI 4.1, OpenAI o3, OpenAI o4-mini, and Grok 3 are available.

## 4.0.14 — March 8, 2025

**Highlights**

- **📝 Canvas** — Edit long-form writing directly inside Bearly. [Docs](https://bearly.ai/docs/features/canvas.md)

**Also new**

- **🖼️ All image engines** — Generate with every engine at once using `/all engines`.

## 4.0.13 — March 2, 2025

**Highlights**

- **🎙️ Speaker-aware transcripts** — Name speakers, follow the current sentence during playback, and let Bearly detect names automatically.

**Also new**

- **⬇️ Media downloads** — Use `/download` to save supported media from the current conversation.
- **🧠 Reasoning steps** — See reasoning-model thoughts in the agent steps view.

**Fixes & improvements**

- GPT-4.5 is available.

## 4.0.12 — February 27, 2025

**Also new**

- **🔁 Rerun** — Generate another answer from a dedicated button.

**Fixes & improvements**

- Claude 3.7 and Claude 3.7 Reasoning are available.
- The input bar has a clearer design.

## 4.0.11 — February 19, 2025

**Fixes & improvements**

- DeepSeek R1 moved to a new hosting provider.
- Google Imagen and Recraft are available for image generation.

## 4.0.10 — January 31, 2025

**Also new**

- **🎞️ Batch transcripts** — Transcribe several recordings together with inline video playback.

**Fixes & improvements**

- OpenAI o3-mini is available with reasoning, tools, and streaming.
- The duplicate code-copy button is fixed.

## 4.0.9 — January 23, 2025

**Also new**

- **🎛️ Model controls** — Set model temperature from Settings.

**Fixes & improvements**

- DeepSeek R1, Gemini 2, and the vision-capable Grok model are available.
- The web-search toggle is back.
- The app loads faster.

## 4.0.8-Beta — December 17, 2024

**Also new**

- **⏹️ Stop responses** — End an answer while it is still generating.

**Fixes & improvements**

- Claude Haiku was upgraded to 3.5.
- Custom prompts can run without fields.
- PDF text is easier to select and highlight.
- Pasted text from Word and other rich documents is handled more reliably.

## 4.0.7-Beta — October 30, 2024

**Highlights**

- **🎤 Turn answers into audio** — Ask Bearly to read something aloud or use `/tts` on a text snippet.

**Also new**

- **📁 Collection search** — Search a specific document collection by typing `/` followed by its name.
- **📸 PDF source previews** — Jump directly from an answer to the cited place in a PDF.
- **🪄 Grammar correction** — Open the writing corrector with `/grammar`.

**Fixes & improvements**

- Document and attachment search includes clearer inline sources.
- xAI Grok is available.

## 4.0.6-Beta — October 22, 2024

**Highlights**

- **🔎 Web search with sources** — Bearly can search the web and show where its answer came from.
- **🫙 Custom prompts** — Create reusable prompts and run them from Focus with `/`.

**Fixes & improvements**

- Claude Sonnet 3.5 was upgraded to its latest release.

## 4.0.5-Beta — October 7, 2024

**Highlights**

- **🎙️ Fast audio transcripts** — Type `/transcript` to turn audio into text.

**Also new**

- **⌨️ Slash commands** — Run tools and shortcuts from Focus with `/`.
- **🔁 Answer versions** — Rerun a question with another model and switch between its answers.

**Fixes & improvements**

- Flux 1.1 image models are available through `/flux`.

## 4.0.4-Beta — September 16, 2024

**Highlights**

- **📋 Switch chats from the sidebar** — Move between conversations directly or through the command menu.

**Fixes & improvements**

- Documents are detected more reliably.

## 4.0.3-Beta — September 12, 2024

**Also new**

- **⬇️ Chat export** — Download a conversation as Markdown.

**Fixes & improvements**

- OpenAI o1 and o1-mini are available.

## 4.0.2-Beta — July 19, 2024

**Fixes & improvements**

- The model picker now includes GPT-4o Mini, GPT-4o, Claude Haiku, and Claude Sonnet 3.5.
- Web search returns better results.
- The global shortcut opens Focus or your most recent notebook mode.
- Answers are less chatty by default.
- Chat titles, the usage meter, and several interface details work more reliably.

## 4.0.1-Beta — June 4, 2024

**Fixes & improvements**

- Code-copy controls now stay out of the way until you need them.
- Chats keep more useful conversation history.
- Chat-switching shortcuts no longer conflict with common system shortcuts.
- Longer pasted text stays in the composer before becoming an attachment.
- Mobile layout, line breaks, and cross-device sync are more reliable.

---

---
title: "Privacy Policy"
description: "Bearly AI Privacy Policy"
canonical: "https://bearly.ai/privacy-policy/"
markdown: "https://bearly.ai/privacy-policy.md"
---

# Privacy Policy

HTML: [https://bearly.ai/privacy-policy/](https://bearly.ai/privacy-policy/)

> Bearly AI Privacy Policy

Last updated: September 25, 2026

## Intro

Privacy is a fundamental human right. We will uphold the privacy of our users by fighting against ads, trackers and other impingements on privacy. We will never profit by monetizing your data. This Privacy Policy explains how Bearly AI (bearly.ai) collects, uses, shares, and retains personal information when you use our websites, apps, and related services (the "Services"). Personal information includes information that identifies or can reasonably be linked to a person, directly or indirectly, including information about other people contained in the material you use with our Services. Third-party services you connect to or interact with have their own privacy policies.

## Information Gathering

We learn information about you when: You directly provide it to us. For example, we collect: Name and contact information. We collect details such as name and email address. Payment information. If you make a purchase, we collect credit card numbers, financial account information and other payment details. Content and files. We collect and retain the photos, documents or other files you send to us in connection with delivering our Services, including via email or chat. We collect it automatically through our products and services. For instance, we collect: Identifiers and device information. When you visit our websites, our web servers log your Internet Protocol (IP) address and information about your device, including device identifiers, device type, operating system, browser, and other software including type, version, language, settings, and configuration. Geolocation data. Depending on your device and app settings, we collect geolocation data when you use our Services. Usage data. We log your activity on our website, including the URL of the website from which you came to our site, pages you viewed on our website, how long you spent on a page, access times, and other details about your use of and actions on our website. We also collect information about which web-elements or objects you interact with on our Service, metadata about your activity on the Service, changes in your user state, and the duration of your use of our Service. Someone else tells us information about you. Third-party sources include, for example: Third-party partners. Third-party applications and services, including social networks you choose to connect with or interact with through our services. Service providers. Third parties that collect or provide data in connection with work they do on our behalf, for example companies that determine your device's location based on its IP address. When we try and understand more about you based on information you've given to us. We infer new information from other data we collect, including using automated means to generate information about your likely preferences or other characteristics ("inferences"). For example, we infer your general geographic location based on your IP address.

### Agents, external tools, and connected accounts

When you use enabled tools, Bearly may access and process information needed to carry out your task. Depending on the tool and platform, this can include file contents and paths, commands and their output, screenshots, text and controls in apps and websites, browser tabs, and records returned by connected services. Screen and browser observations can include information about other people or unrelated sensitive material visible in the selected app or display. Review the access you enable and close or hide material you do not want included.

Relevant task information may pass through Bearly's systems and be sent to AI providers processing your request. Tool calls may also send information to the website, connected account, or external server involved in the task. Running a tool locally does not mean that all information it reads stays on your device. Tool results and activity records may become part of your Bearly chat and are then handled like other chat content under this policy.

Tools can act through accounts you have signed in to. Bearly Browser uses its own browser session; importing browser cookies is a separate choice that can give it access to existing sign-ins. Connected Chrome uses your connected Chrome profile, including signed-in sites. Connector sign-in and configuration may involve access tokens or other authentication information needed to reach the service. Connecting an account and granting operating-system permissions remain separate from choosing a chat's or routine's permission mode.

### Permission choices and organization controls

In an interactive chat, **Standard** asks for tool approval when needed and honors saved approvals. Where available, **Full access** lets enabled external tools run without a separate Bearly approval for each action, including actions that read, change, delete, or send information. For an interactive chat, it applies to that chat on one device. Tools and accounts you deliberately enable later can be used under that chat's mode. Full access changes when Bearly asks for approval; it does not change the purposes for which we use personal information under this policy or remove your privacy rights.

Bearly remembers the permission choice for your account and chat on the current device. Restarting the app or returning after 24 hours away pauses Full access in that chat. You can choose **Standard**, revoke saved approvals, or use **Clear all permissions** from the chat's permission control. Turning off Full access preserves saved approvals unless you also revoke them. These controls do not undo actions already completed or delete information already saved in a chat or received by another service. You can also disconnect a service in Bearly and revoke its authorization at that service. See [Choose how Bearly uses external tools](https://bearly.ai/docs/settings/agent-permissions.md) for the available controls.

Routines require separate Full access approval on the desktop device that runs them. Bearly keeps a device-local record of that approval, associated with the account, device, routine and the saved instructions and selected tool setup. It survives app restarts and time away; each run still requires current authorization. Material changes to the saved routine or its selected connector and skill setup require a new review. Copies and routines moved to another device start in Standard. Choosing Standard for a routine stops its Full access while preserving existing per-tool unattended permissions. A change made on another device takes effect when it syncs to the running device. Routine tool results and activity may be retained in the routine's chat like other chat content.

For managed accounts, organization administrators control eligibility for Full access; members still choose whether to enable it for a chat or separately for a routine. We keep administrative audit records of eligibility changes, including the administrator, affected member or organization, time, and changed setting. Authorized organization administrators can review those changes. Granting eligibility does not itself give an administrator access to a member's private chat content. Turning off Full access does not delete administrative audit records; their retention is governed by the purposes and legal obligations described below.

We also process permission-mode changes, approval decision sources, authorization failures, and related account or device identifiers to enforce access controls, investigate problems, and maintain service security and reliability. Full Access-specific access telemetry does not include the contents of commands, files, screenshots, messages, or credentials. This is separate from task information processed by tools or stored in chats as described above.

## Information Use

We use each category of personal information about you:

- To provide you with our Services
- To improve and develop our Services
- To communicate with you
- To provide customer support

### Google API data

If you connect Google Workspace, Bearly uses the Google account data you authorize only to provide the Gmail and Google Calendar actions you request. Gmail actions are limited to searching and reading email, messages, and threads; listing drafts and labels; creating drafts; removing the Inbox label to archive a message or thread; and moving a message or thread to Gmail Trash. Bearly does not send or permanently delete email, make other label changes, or download attachment files through this connection. Bearly receives attachment names, types, and sizes when Gmail provides them. You can find archived content in Gmail's All Mail view and move it back to Inbox. You can restore content from Gmail Trash until Google permanently removes it under Gmail's retention rules.

Google Calendar actions are limited to listing calendars; reading events in a time range; and creating, updating, or deleting timed, one-time events without guests. Calendar data Bearly receives can include calendar and event identifiers, titles, descriptions, locations, times, time zones, organizer and creator details, attendees and response status, and recurrence information. Bearly does not invite guests, change or delete events with guests, change or delete recurring events, create calendars, or change calendar sharing. A deleted event moves to Google Calendar Trash for 30 days and can be restored there using Google Calendar on a computer.

These actions follow your instructions and the chat's tool permissions. In Standard, Bearly asks when an action is not covered by an applicable saved approval. In Full access, supported actions can proceed without separate action-approval prompts. Full access does not connect a Google account, expand the Google permissions you granted, or add unsupported operations. The operation limits above describe the Google Workspace connector; browser and computer-control tools are separate capabilities with their own access controls.

Relevant Gmail and Calendar content may pass through Bearly's systems and be sent to the AI model you selected to complete your request. Connector results can become part of the Bearly chat where you used them and are then handled like other chat content under this policy. Bearly does not use Google API data for advertising, sell it, or use it to determine creditworthiness or for lending purposes.

You can disconnect or remove Google Workspace in **App Settings** → **Connectors**. You can separately revoke Bearly's Google authorization from your [Google Account connections](https://myaccount.google.com/connections). To delete connector-derived content saved in Bearly, delete the chat that contains it or contact [support@bearly.ai](mailto:support@bearly.ai) for an account deletion request.

Bearly's use and transfer to any other app of information received from Google APIs adheres to the [Google API Services User Data Policy](https://developers.google.com/terms/api-services-user-data-policy), including the Limited Use requirements.

## Information Sharing

We share information about you: When we've asked & received your consent to share it. As needed, including to third-party service providers, to process or provide Services or products to you. To comply with laws or to respond to lawful requests and legal process, provided that we'll notify you unless we're legally prohibited from doing so. We'll only release personal information if we believe in good faith that it's legally required. Only if we reasonably believe it's necessary to prevent harm to the rights, property or safety of you or others. In the event of a corporate restructuring or change in our organizational structure or status to a successor or affiliate. Please note that some of our Services include integrations, references, or links to services provided by third parties whose privacy practices differ from ours. If you provide personal information to any of those third parties, or allow us to share personal information with them, that data is governed by their privacy statements. Finally, we may share non-personal information in accordance with applicable law.

## Information Protection

We implement physical, business and technical security measures to safeguard your personal information. In the event of a security breach, we'll notify you so that you can take appropriate protective steps. We only keep your personal information for as long as is needed to do what we collected it for. After that, we destroy it unless required by law.

## Other Information

We retain personal data for as long as necessary to provide the services and fulfill the transactions you have requested, comply with our legal obligations, resolve disputes, enforce our agreements, and other legitimate and lawful business purposes. Because these needs can vary for different data types in the context of different services, actual retention periods can vary significantly based on criteria such as user expectations or consent, the sensitivity of the data, the availability of automated controls that enable users to delete data, and our legal or contractual obligations. As part of our normal operations, your information may be stored in computers in other countries outside of your home country. By giving us information, you consent to this kind of information transfer. Irrespective of where your information resides, we'll comply with applicable law and abide by our commitments herein. We don't want your personal information if you're under 13. Do not provide it to us. If your child is under 13 and you believe your child has provided us with their personal information, please contact us to have such information removed.

## European Economic Area, United Kingdom, Swiss and California Users

The following rights are granted under the European General Data Protection Regulation ("GDPR") and California Consumer Privacy Act ("CCPA"). Bearly AI applies these rights to all users of our products, regardless of your location: The right to know what personal information is collected. The right to know if personal information is being shared, and to whom. The right to access your personal information. The right to exercise your privacy rights without being discriminated against. EEA, UK, and Swiss Users: Our lawful bases for collecting and processing personal information under the GDPR include: Performing our contract with you and providing our services. Legitimate interests: we receive technical and interaction data of users, which may include IP addresses, to improve the security and reliability of our services and prevent abuse, and to understand where people learn about Bearly AI. Consent: where we ask for your consent to process your information, you can always withdraw this consent. Under the GDPR, EEA, UK, and Swiss users have additional rights: The right to request correction or erasure of personal information. The right to object to processing your personal information. The right to transfer or receive a copy of the personal information in a usable and portable format, when any automated processing of personal data is based on your consent or a contract with you. The right to withdraw your consent to processing, when the processing is based on your consent. When we are processing data on behalf of another party that is the "data controller," you should direct your request to that party. You also have the right to lodge a complaint with a supervisory authority, but we encourage you to first contact us with any questions or concerns. California Users: Under the CCPA, California residents have additional rights: The right to request personal information to be deleted, subject to several exceptions. The right to opt-out of the sale of personal information. Note that we do not "sell" personal information as defined by the CCPA and have not done so in the past 12 months.

## Exercising Your Rights

To exercise the rights described above or if you have a question or concern, please contact us at support@bearly.ai and we will aim to address or resolve it. If we cannot, you have the right to lodge a complaint with your local data protection authority.

## Changes

We may update this Privacy Policy and related notices from time to time. We will identify the updated policy with a date and provide additional notice of material changes, or obtain consent, where required by applicable law. Choosing Full access is a tool-authorization choice, not consent to unrelated uses of personal information.

## Contact Us

You can get in touch by emailing us at support@bearly.ai.

---

---
title: "Terms of Service"
description: "Bearly AI Terms of Service"
canonical: "https://bearly.ai/terms-of-service/"
markdown: "https://bearly.ai/terms-of-service.md"
---

# Terms of Service

HTML: [https://bearly.ai/terms-of-service/](https://bearly.ai/terms-of-service/)

> Bearly AI Terms of Service

Last updated: September 25, 2026

PLEASE READ THESE TERMS OF SERVICE ("AGREEMENT" OR "TERMS OF SERVICE") CAREFULLY BEFORE USING THE SERVICES OFFERED BY BEARLY.AI. THIS AGREEMENT SETS FORTH THE LEGALLY BINDING TERMS AND CONDITIONS FOR YOUR USE OF THE BEARLY.AI WEBSITE AND ALL RELATED SERVICES AND APPLICATIONS, INCLUDING, WITHOUT LIMITATION, ANY FEATURES, CONTENT, WEBSITES (INCLUDING WWW.BEARLY.AI) OR APPLICATIONS OFFERED FROM TIME TO TIME BY Bearly AI IN CONNECTION THEREWITH (COLLECTIVELY "SERVICE(S)"). BY USING THE SERVICES IN ANY MANNER, YOU AGREE TO BE BOUND BY THIS AGREEMENT.

## Acceptance of Terms of Service

The Service is offered subject to acceptance without modification of all of these Terms of Service and all other operating rules, policies and procedures that may be published from time to time in connection with the Services by Bearly AI. In addition, some services offered through the Service may be subject to additional terms and conditions promulgated by Bearly AI from time to time; your use of such services is subject to those additional terms and conditions, which are incorporated into these Terms of Service by this reference. Bearly AI may, in its sole discretion, refuse to offer the Service to any person or entity and change its eligibility criteria at any time. This provision is void where prohibited by law and the right to access the Service is revoked in such jurisdictions.

## Modification of Terms of Service

At its sole discretion, Bearly AI may modify or replace any of the Terms of Service, or change, suspend, or discontinue the Service (including without limitation, the availability of any feature, database, or content) at any time by posting a notice on the Bearly AI websites or Service or by sending you an email. Bearly AI may also impose limits on certain features and services or restrict your access to parts or all of the Service without notice or liability. It is your responsibility to check the Terms of Service periodically for changes. Your continued use of the Service following the posting of any changes to the Terms of Service constitutes acceptance of those changes.

We will provide any additional notice and obtain any agreement required by applicable law for material changes to these Terms.

## Rules and Conduct

As a condition of use, you promise not to use the Service for any purpose that is prohibited by the Terms of Service. For purposes of the Terms of Service, the term "Content" includes, without limitation, any information, data, text, photographs, videos, software, scripts, graphics, and interactive features generated, provided, or otherwise made accessible on or through the Service.

## Agents, external tools, and Full access

Bearly's agents can use enabled tools to carry out your instructions outside the chat. Depending on the tool and platform, this can include reading, changing, or deleting files; running programs; observing and controlling a computer or browser; and taking actions in connected accounts or signed-in apps and websites. These actions can disclose information, send communications, create commitments, incur third-party charges, or change external records.

**Your authorization.** You authorize Bearly to use enabled tools to carry out your instructions within your chosen permission settings and applicable access restrictions. You must have the rights and permissions needed for the accounts, systems, content, and actions you direct Bearly to use, including any required authority or consent concerning other people's information. You remain responsible for the instructions you provide, the access you grant, and complying with applicable law and third-party terms. If you act for an organization, you must have authority to do so.

Organization administrators must be authorized to manage members' eligibility and must provide any notices or obtain any permissions required by their organization and applicable law. Allowing eligibility does not replace a member's explicit choice to activate Full access.

**Permission modes.** In interactive chats, Standard asks for approval when needed and honors saved approvals. Where available, selecting Full access authorizes enabled external tools to perform actions for your tasks without a separate Bearly approval prompt for each action. For a chat, Full access applies to one chat on one device and includes tools or accounts you deliberately enable later. You must explicitly activate it; organization eligibility alone does not activate a member's chat. Restarting Bearly or returning after 24 hours away pauses Full access in that chat until you choose to resume.

**Scheduled routines.** Full access for a routine requires separate explicit approval on its installed desktop device. This authorizes unattended actions through that routine's selected tools and accounts for its saved instructions. The approval survives app restarts and time away, but each run still requires current authorization. Material changes to saved instructions, scheduling, delivery or selected tool and account setup require renewed review. Copying or moving a routine does not transfer Full access approval. Standard routines retain their existing, separately enabled unattended permissions. Full access does not add operations or tools unavailable to Routines.

Full access does not override organization rules, operating-system permissions, required safety checks, account permissions, tool limitations, or applicable law. It does not itself authorize connecting another account or setting up a disabled tool. A selected working folder provides context and does not confine access to that folder. Tool availability differs by device, and choosing Full access on a phone does not give it desktop capabilities or activate access on another device. Scheduled Routines keep their own separate authorization requirements.

**Risks and review.** AI agents can misunderstand instructions, make incorrect changes, or be misled by content in files, websites, messages, or tool results. Full access reduces opportunities to catch mistakes before an action occurs. Review the tools and accounts you enable, monitor consequential work, and check results before relying on them. Use Standard in an interactive chat when you want action-approval prompts. Bearly does not guarantee that an agent will recognize every risk or that an action will succeed, be accurate, or be reversible.

**Stopping and revoking access.** You can stop a task or switch the chat or routine to Standard. Stop does not change the selected permission mode or a routine's schedule. Switching to Standard preserves saved approvals, including separately enabled unattended routine permissions; Clear all permissions in a chat also removes that chat's saved approvals. Changes to a routine made on another device take effect when they sync to its running device. Organization administrators can withdraw eligibility for managed members, but group administrators cannot grant it. Cancellation and revocation cannot undo completed actions or guarantee recall of requests already sent to an external service or work that continues outside Bearly's control. Check the affected system before retrying an interrupted action. See [Choose how Bearly uses external tools](https://bearly.ai/docs/settings/agent-permissions.md) for the controls and their limits.

**Privacy and third parties.** Information retrieved or generated by tools may be processed by Bearly, AI providers, and the external services involved in the task, as described in our [Privacy Policy](https://bearly.ai/privacy-policy.md). Third-party services remain subject to their own terms and privacy practices. Full access is not a waiver of your privacy rights or consent to unrelated data use.

## Age Requirements

Use of Bearly AI by anyone under 18 years of age is prohibited. You represent that you are at least the age of majority in the jurisdiction where you live or, if you are not, your parent or legal guardian must consent to these Bearly AI Terms of Service and affirm that they accept these Terms on your behalf and bear responsibility for your use.

## Restricted Uses

You agree NOT to use Bearly AI to:

- Violate the rights of another party or any applicable laws
- Reverse assemble, reverse compile, decompile, translate or otherwise attempt to discover the source code or underlying components of models, algorithms, and systems of Bearly AI, or their underlying technologies
- Abuse, harm, interfere with, reverse engineer, or disrupt Bearly AI, or their underlying technologies
- Use unauthorized automation to access the Service, bypass access controls, or exceed applicable rate limits. Supported Bearly agents, tools, and Routines may be used within their available controls and these Terms
- Access, extract, or disclose data you are not authorized to access, or bypass the Service's protections to do so
- Represent that Bearly AI Content is human-generated

## Trademarks and Patents

All Bearly AI logos, marks and designations are trademarks or registered trademarks of Bearly AI. All other trademarks mentioned in this website are the property of their respective owners.

## Termination

Bearly AI may terminate your access to all or any part of the Service at any time if you fail to comply with these Terms of Service, which may result in the forfeiture and destruction of all information associated with your account. Any fees paid hereunder are non-refundable. Upon any termination, all rights and licenses granted to you in this Agreement shall immediately terminate.

## Indemnification

You shall defend, indemnify, and hold harmless Bearly AI, its affiliates and each of its, and its affiliates employees, contractors, directors, suppliers and representatives from all liabilities, losses, claims, and expenses, including reasonable attorneys' fees, that arise from or relate to (i) your use or misuse of, or access to, the Service, or (ii) your violation of the Terms of Service or any applicable law, contract, policy, regulation or other obligation.

## Limitation of Liability

IN NO EVENT SHALL Bearly AI OR ITS DIRECTORS, EMPLOYEES, AGENTS, PARTNERS, SUPPLIERS OR CONTENT PROVIDERS, BE LIABLE UNDER CONTRACT, TORT, STRICT LIABILITY, NEGLIGENCE OR ANY OTHER LEGAL OR EQUITABLE THEORY WITH RESPECT TO THE SERVICE FOR ANY LOST PROFITS, DATA LOSS, COST OF PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES, OR SPECIAL, INDIRECT, INCIDENTAL, PUNITIVE, OR CONSEQUENTIAL DAMAGES OF ANY KIND WHATSOEVER.

## Disclaimer

ALL USE OF THE SERVICE AND ANY CONTENT IS UNDERTAKEN ENTIRELY AT YOUR OWN RISK. THE SERVICE IS PROVIDED "AS IS" AND "AS AVAILABLE" AND IS WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED.

Nothing in these Terms, including the Full access acknowledgment, indemnification, limitations of liability, or disclaimer, excludes or limits rights, remedies, or liability that cannot lawfully be excluded or limited under applicable law.

## Miscellaneous

The Terms of Service are the entire agreement between you and Bearly AI with respect to the Service, and supersede all prior or contemporaneous communications and proposals between you and Bearly AI with respect to the Service.

## Contact Us

For questions regarding the Service, you can get in touch by emailing us at support@bearly.ai.

---