Glossary
Every term below is written the way the app shows it. Each definition links to the manual page with the details.
Symbols
Section titled “Symbols”/ command: Type / at the very start of a message to pick a tool, research feature, skill, or prompt template by name, such as /deep-research or /chart. A message can start with one / command. See Write, attach, and send.
@ reference: Type @ anywhere in a message to pick a file, tool, or skill by name. It becomes a chip that tells the AI exactly what you mean, and a message can hold any number of them. See Write, attach, and send.
Account menu: The row at the bottom of the sidebar showing your email, plan, and usage meter. It opens App Settings, Billing & Pro, Organizations, Give Feedback, and Logout. See Tour the workspace.
Agent access: A managed organization’s admin setting that decides which members may turn on Full access. See Organizations.
Agentic (Compaction): A History setting for long chats. When a conversation gets long, Bearly writes a summary of the older part and keeps recent messages word for word, so long tool-heavy work can keep going. See Tune a chat.
Always Search Documents: A project chat setting that makes every answer in the project search the project’s files first. It has no effect in end-to-end encrypted chats. See Documents.
Analyst: A paid plan above Pro that adds Deep Research (Standard) and People Search. See Plans, credits, and limits.
Approval duration: How long an approval lasts: Once, This response (until the answer ends), This session (until the chat reloads), or This chat (on this device until you revoke it). Approvals never sync to your other devices. See Permissions.
Archive: Hides a chat from the main list without deleting it. Archives at the bottom of the sidebar lists archived chats, and an archived chat must be unarchived before it can be deleted. See Organize chats.
Ask Experts: The name /council shows in the / menu. See Council of Experts.
Assets: The heading in the / and @ menu that lists the files in this chat and its project. See Write, attach, and send.
Bearly AI Models: A built-in skill that lets scripts the AI writes call Bearly’s AI models. It works on every platform. See Connectors and skills.
Bearly Browser: A separate browser inside the desktop app that chats can use in background tabs, with its own site sign-ins. Bearly asks before it uses a new site unless the chat has Full access. See Browser and computer use.
Bearly Fast, Bearly Pro, Bearly Ultra: Bearly’s own model choices, described in the model picker as always kept up to date. New chats on Free default to Bearly Fast. Paid plans default to Bearly Pro, or to Bearly Fast once you’ve used 95% or more of your allowance. See Choose and compare models.
Best Pick: Researches expert reviews, store listings, and owner reviews, then recommends exactly one product to buy, with a plain store link. Pro and higher. See Research.
Browse files: The button in the chat header that lists the chat’s files under Uploaded and Generated, plus Project in a project chat. See Documents.
Browser extension: Bearly’s extension for Chrome, where it’s listed as Bearly Companion, and Firefox. Its side panel summarizes, explains, and answers questions about the page you’re on. In Chrome, it also lets the desktop app control your tabs. See Get Bearly.
Build on previous runs: A routine setting that lets each run see earlier runs in the routine’s chat. It’s on by default. See Routines.
Building block: This manual’s word for a capability seen as a step: what it takes in, and what it leaves in the chat for the next step. See Building blocks.
Canvas: An editable document made from an answer with Edit in Canvas. Select text and ask the AI to change it, or edit the Markdown yourself. See Pages and Canvas.
Chapters: Short labels Bearly adds about every 5 turns to name stretches of a conversation. They show as Chapter ticks on the timeline rail. See Work with answers.
Chat (Dynamic): The default History setting. The AI sees the newest messages in full and shorter excerpts of older ones. See Tune a chat.
Chat management: Lets the AI list, read, create, rename, move, and delete your chats. It asks before listing, reading, or deleting, even with Full access. See Organize chats.
Chrome control: In the desktop app, Bearly can read and act in your regular Chrome tabs through the Bearly extension, with your approval. See Browser and computer use.
Command menu: Opens with ⌘ K (Ctrl K on Windows and Linux) or … in the chat header. It holds chat actions such as New, Rename, Download as Markdown, and New Routine, and a list for switching chats. See Organize chats.
Composer: The message box at the bottom of a chat, with its toolbar and the context strip under it. See Write, attach, and send.
Computer Use: In the desktop app, Bearly can look at and control the apps open on your screen to finish a task you approve. Computer use runs on your device; the screenshots you approve go to the model you picked. See Browser and computer use.
Connectors: Services and MCP servers you connect so chats and routines can use their tools. A connector works in a chat only after you turn it on there. See Connectors and skills.
Context strip: The row of chips under the message box: project, encryption lock, Work locally, Connectors, and Permissions. See Tour the workspace.
Council of Experts: /council sends one question to 2 to 5 models at once, shows each answer, then combines them into one reply. Pro and higher. See Research.
Cowork: Another name some screens use for Work locally.
Credit Boost: One-time credits you can buy on a paid plan to add to this month’s allowance. They expire at the end of the current month, or of your seat’s period. Boosts aren’t sold inside the iOS app. See Plans, credits, and limits.
Credits: The AI allowance that comes with your plan. Models and paid tools draw from it, and the usage meter shows how much you’ve used. See Plans, credits, and limits.
Deep Research: A long research run that produces a cited report after you review and edit its plan. The report shows in the chat, and its PDF is saved there and emailed to you. Standard needs Analyst; Ultra runs about twice as deep and needs Research. See Research.
Deep Summary: The summary Summarize Doc (/summarize) writes for a long document. See Documents.
Desktop app: Bearly for Mac, Windows, and Linux. It adds Work locally, Computer Use, Bearly Browser, Chrome control, installed skills, running routines, and, on Windows and macOS, recording a meeting’s computer audio. See Get Bearly.
Devices: The settings pane that lists where your account is signed in, with Sign out for each device and the keys used by the browser extension and the API. See Privacy and security.
Dictation: Speech to text in the message box. Click the microphone, or hold the push-to-talk key (Right Shift by default) while Bearly is focused. It’s off until you turn on Enable dictation. See Voice.
Docs & Data: A project’s files page. Files here are available to every chat in the project. See Projects.
Document editor: /document creates and edits PDF, Word, PowerPoint, and Excel files, including filling PDF forms and adding tracked changes to .docx files. See Documents.
Edit in Canvas: The button under an answer that opens it as a Canvas.
End-to-end encryption: An opt-in setting, labeled Beta, for new chats. Turn on end-to-end encryption and your new chats are encrypted on your devices. Bearly stores and syncs only the locked copies, and each message still goes to the model to be answered. See Privacy and security.
Free: The plan with no charge. It includes a small daily and monthly allowance, and new chats use Bearly Fast. See Plans, credits, and limits.
Full access: A permission mode for one chat on one device that lets Bearly use your enabled tools without asking each time. It still asks before connecting a new service and before listing, reading, or deleting your chats. See Permissions.
Generated: The Browse files tab for files Bearly made in the chat, such as Pages, CSV files, images, and audio. See Documents.
Get Video: /download-video saves videos from sites such as X as MP4 files in the chat. YouTube links work in the desktop app only. See Videos and posts.
History: A per-chat setting in the sliders button that decides how older messages are kept: Chat (Dynamic) or Agentic (Compaction). See Tune a chat.
Imagine: /image creates images from a description. Results open in an inline image editor. See Images, video, and audio.
Launcher: The screen a new chat opens on, with a headline, the message box, a project chip, and daily tips. See Tour the workspace.
Local Directory Sync: A project setting in the desktop app that keeps a folder on your computer synced into the project’s files. See Projects.
Local key encryption: The lock label on new chats when your organization’s policy requires local key encryption. You can’t switch it off for those chats. See Privacy and security.
Managed organization: An organization with an email domain Bearly has set up for it. It unlocks policies, usage reports, spending limits, and Agent access. See Organizations.
Max: The highest plan, sold only as organization seats. See Plans, credits, and limits.
MCP server: A server that offers tools to AI apps. Add your own as a custom connector, over HTTPS or, in the desktop app, as an npm package. See Connectors and skills.
Meetings and Recordings: The sidebar page for recording a meeting or voice note, with live and final transcripts, speaker names, and summaries on request. See Meetings.
Memories: A chat setting that lets the AI search your past chats with Recall when you refer to earlier work. In a project chat you can limit it to Project Only. It isn’t available in end-to-end encrypted chats. See Tune a chat.
Model picker: The button at the right of the toolbar that shows the current model and chooses which model answers in this chat. See Choose and compare models.
Needs attention: A sidebar chip with a count of other chats that are waiting for your approval or have replies you haven’t read. See Organize chats.
New Chats: The settings pane for the defaults new chats start with. Project chats use project settings instead. See Tune a chat.
Notification filter: A routine condition, such as “A competitor ships something new”. Every run still lands in the routine’s chat; Bearly compares each result with the last one and sends your email or text only when the run meets the condition, fails, or it can’t tell, plus a check-in every seventh quiet run. See Routines.
Organization: A group of members with shared groups, projects, and billing, plus policies in a managed organization. Open it from the account menu → Organizations. See Organizations.
Page: A single-file interactive app, such as a calculator, dashboard, game, or planner, that Bearly writes as a file and runs live in the chat. You can publish a Page to a stable bearly.sh link. See Pages and Canvas.
People Search: Identifies one named person after you confirm the right candidate, then builds a cited brief. It also finds people from a description and looks up work emails. Analyst and higher. See Research.
Permissions: The composer control for this chat’s permission mode, Standard or Full access, and its saved approvals. See Permissions.
Personalization: Context you write about yourself (how to address you, your work, anything else useful), added to your chats only while you have the switch on. See Tune a chat.
PII Blocker: An organization setting that replaces email addresses, phone numbers, credit card numbers, and US Social Security numbers in your messages with placeholders before they’re sent. See Write, attach, and send.
Plan: Your subscription level: Free, Pro, Analyst, Research, or Max. Higher plans add credits and features. See Plans, credits, and limits.
Policy: A managed organization admin’s set of rules for members, covering which models and features they can use and how much they can spend. See Organizations.
Pro: The first paid plan. It adds Routines, Council of Experts, Best Pick, and the choice of voice-call model. See Plans, credits, and limits.
Project: A folder for chats that shares files (Docs & Data), instructions, and default settings. You can share a project with people, groups, or a whole organization, and protect it with a password. See Projects.
Prompt Template: A saved prompt with fill-in fields. Pick it from the / menu under Skills, fill in the form, and select Submit; the filled form stays in the chat so you can run it again. See Write, attach, and send.
Queue: Messages you send while Bearly is still answering wait under Up next and go out in order. It holds up to 4 and pauses after an error or at a usage limit. See Write, attach, and send.
Read Aloud: Turns text into speech. /say makes an audio file and opens Voice Studio; the Read aloud button under a text answer plays it. See Images, video, and audio.
Read Site: /read-site reads one web page. Long pages, documents, images, and media it finds are saved as files in the chat. See Research.
Read-only chat: A chat you can view but not write in, such as one in a project where you can only read or one in a locked project. It has no message box. See Projects.
Reasoning Effort: How much a model thinks before it answers: Default, None, Low, Medium, High, or Extra High. How much each level changes depends on the model. See Tune a chat.
Recall: The tool Memories uses to search your past chats. While Memories is on, type /recall to make it search now. See Tune a chat.
Recommended: The New Chats model choice that follows Bearly’s default: Bearly Fast on Free, Bearly Pro on paid plans until you’ve used 95% of your allowance, then Bearly Fast. See Choose and compare models.
Recovery phrase: The 24 words you save when you set up end-to-end encryption. It opens your encrypted chats on a new device. Lose every approved device, your recovery phrase, and any saved passkey, and nobody can recover your encrypted chats, including Bearly, so save it. See Privacy and security.
Research: A paid plan above Analyst that adds Deep Research Ultra. See Plans, credits, and limits.
Response Style: How answers are written: Normal, Classic, No Style, Learning, Concise, Explanatory, or Formal. See Tune a chat.
Routine: A saved prompt that runs on a schedule in its own chat, and can email or text you the result. Pro and higher; routines run in the desktop app, on the device they’re installed on, while it’s open. See Routines.
Run Python: /python runs Python code in a fresh cloud sandbox on files you name and returns results, plots, and up to 5 files. See Data and code.
Saved approvals: Approvals you gave for longer than once in a chat. Review and revoke them in Permissions. See Permissions.
Scholar: Searches academic papers, reads passages, and cites them with superscript markers. It runs when Scholar is Auto, or when you type /scholar. See Research.
Search chats and files: Finds messages and files across your chats by meaning and exact words. Open it with ⌘ F (Ctrl F on Windows and Linux) or the sidebar magnifier. It leaves out end-to-end encrypted chats. See Organize chats.
Share: Turns the substance of a conversation, or one file, into a Page with artwork and a link preview. You review it, then publish it to a link. See Pages and Canvas.
Skill: A packaged set of instructions, with optional files and scripts, that Bearly loads when a task calls for it. Install skills in the desktop app from Settings → Skills; Visualize and Bearly AI Models are built in. See Connectors and skills.
Skill Store: The section of Settings → Skills in the desktop app for finding and installing skills. See Connectors and skills.
Standard (permissions): The default permission mode: Bearly asks for approval when a tool needs it, and approvals you saved still apply. See Permissions.
Standard encryption: The lock label on chats that aren’t end-to-end encrypted, shown when end-to-end encryption is available to you. Chats and files are encrypted at rest, each with its own key. See Privacy and security.
Standard retention: What the model picker’s Standard badge means: the model’s main route has no verified zero-data-retention guarantee. See Choose and compare models.
Temperature: How varied answers are, from 0 to 2: Factual (0.5), Default (1), or Creative (1.5). Above 1.5, output can become incoherent. See Tune a chat.
Thinking: Collapsible steps in an answer that show a model’s reasoning. See Work with answers.
This Chat: The settings pane for the open conversation only. See Tune a chat.
Timeline rail: A row of ticks on the right edge of wide windows, one per prompt, for jumping around a long chat. See Work with answers.
Touch Up: /edit-image edits, extends, or combines images in the chat. See Images, video, and audio.
Transcribe: /transcribe turns audio and video files into a transcript with speaker names and timestamps. See Images, video, and audio.
Unattended local access: A routine setting, turned on only in the desktop app, that lets scheduled runs use Work locally and approved skill scripts without waiting for you. It asks you to confirm the risks first. See Routines.
Up next: The queue of messages waiting to send. See Queue.
Usage meter: The bar in the account menu showing how much of your allowance you’ve used. See Plans, credits, and limits.
Version switcher: The ‹ 2 / 3 › arrows under a message you edited or reran, for flipping between versions. See Work with answers.
View Media: /analyze-media lets the AI look at image and video files in the chat, up to 10 minutes of video at a time. See Images, video, and audio.
Visualize: A built-in skill, /visualize, for interactive Pages, charts, and diagrams. See Data and code.
Voice call: A live spoken conversation that can use the chat’s tools. Start it with Start voice call in the composer; the transcript is saved in the chat, the audio isn’t. See Voice.
Voice Studio: The panel on your latest text-to-speech result for choosing a voice, a delivery, and a preset, then making and comparing takes. /say with no text opens it. See Images, video, and audio.
Web: A per-chat setting in the sliders button, Off or Auto. On Auto, Bearly searches and reads the web when it helps. See Tune a chat.
Work locally: In the desktop app, lets a chat read, write, and edit files and run commands in a folder you choose. The folder is working context, not a sandbox. See Browser and computer use.
ZDR: Short for zero data retention. In the model picker, the ZDR badge marks zero-retention models. See Choose and compare models.
Zero Data Retention only: An account setting in Settings → Privacy & Security. Turn on Zero Data Retention only and Bearly shows only zero-retention models, and automatic fallbacks stay on them too. It applies to chat models. See Privacy and security.