Capture knowledge
Every way to get something into Breno: files, notes, links, voice, photos, meetings, the browser extension, and connected tools.
Everything you capture becomes an item in your workspace's Knowledge, read and indexed in the background and searchable once it has been read. This page covers each way in, and what to expect.
From the app#
Open Knowledge and choose Add to knowledge. The menu offers:
| Choice | What it does |
|---|---|
| Upload files | Adds one or more files. You can also drag files onto the page. |
| Add link | Saves a web page, video or file link. Breno fetches and reads it in the background. |
| Add text | Saves a note with any title you give it, as Markdown. Notes stay editable. |
| Record voice | Records audio in your browser, transcribes it, lets you edit the text, and saves it. |
| Photo | Reads a photo (for example of a whiteboard), lets you edit what it read, and saves it. |
| Meeting | Opens Meetings, where you turn a transcript or recording into a summary, decisions and action items. |
Files#
Upload PDFs (text and scanned), Word documents, PowerPoint decks, Excel and CSV files, HTML, plain text, Markdown, images (PNG, JPG, WEBP) and audio (M4A, MP3, WAV, OGG). Any Unicode filename works. The limit is 25 MB per file, and the limits are shown in the upload dialog. The full list, and what is not supported, is in Supported files and limits.
Adding a document costs coins: one coin per page started. Importing many files at once shows the estimated cost first and asks you to confirm. See Billing and coins.
Links#
Paste a web address. Breno reads articles, YouTube videos and podcasts, and extracts PDF, Office, CSV, image and audio links by their type. Addresses are normalized, so the same page pasted twice is recognized.
A link Breno cannot read (a type it does not support) is refused immediately with a clear reason, and nothing is saved. Pages that only render with JavaScript are not supported yet.
Notes#
Add text gives you a Markdown note. Edit it any time; each edit keeps the version before it. See Notes and versions.
The browser extension#
The extension captures a text selection, the page address, or a screenshot into your workspace. It acts with a token scoped to one workspace. Screenshots need a vision provider.
Through MCP and the API#
An AI tool connected over MCP can save notes and links with capture_note and capture_url, and the REST API can upload files, capture links and edit your own notes. These need a token with the write scope, and viewers cannot write. See Write and MCP server.
From connected tools#
Connect Slack, Google Drive, Gmail, Notion, GitHub, Jira or Linear and Breno brings their content in and keeps it in step. See Connect sources.
What happens after you capture#
- Read. Breno extracts the text, using OCR for scanned pages and transcription for audio.
- Index. The text is split into passages and embedded, so it can be found by meaning and by exact words. An item is searchable even if the later enrichment step fails, and it is flagged when that happens.
- Understand. Breno recognizes entities and proposes connections for your review. See Memory and decisions.
The item shows its status. A failed item shows a reason, and Retry runs the whole path again; temporary failures are retried automatically with backoff.
Duplicates#
Identical content added to the same workspace is recognized: Breno returns the existing item and tells you it is Already in your knowledge. The same content in another workspace is a new item, because workspaces never see each other.
Who owns what you add#
Every item has an owner: the person who added it, or for a synced item the person who made the connection. You can make an item Only me, and set its sensitivity and space from the item page. See Organize knowledge.