# Memory: documents, canvases and files that write themselves with you.

Source: https://www.secondbrain.media/guide/memory  
Language: en  
Spanish version: https://www.secondbrain.media/guide/memory.es.md

---

This is the part of the workspace where knowledge stops being a daily note and becomes something you consult. A document tree with a full editor, real-time collaboration, Miro-style canvases with Apple Pencil, files with inline preview, and an AI that generates and rewrites using your own material as context.

## The document tree

Documents are organized in a tree of folders and sub-documents, in the order you give them — by dragging — rather than the one a date imposes.

- Free nesting: a document can have children, and those children can have their own.
- Moving a document takes its whole branch with it.
- Trash with restore: nothing is deleted in one shot.
- A document can be moved to another workspace without losing its associations.

_The tree on the left, the document on the right. Under the title sits its parent: that is where it hangs, and how it moves._

## The editor

It is the same editor as the daily note, with all the formatting a real document needs.

- **Full formatting** — Headings, lists, checkboxes, tables, quotes, syntax-highlighted code, images and attachments. Any block can be dragged to reorder it.
- **Markdown, Mermaid and LaTeX** — Write in Markdown if that is faster for you, draw diagrams with Mermaid and set formulas with LaTeX. It all renders in the document.
- **Knowledge intellisense** — As you type, autocomplete people, projects and organizations from your Atlas. Every mention is linked, and the document shows up on that entity's entry.
- **Live attachments** — Files you upload to a document stay attached to it, preview inline, and their text enters search.

## Real-time collaboration

Two people writing in the same paragraph at the same time do not clobber each other: editing is genuinely concurrent, with live cursors and presence, and the document converges with no conflicts and no duplicate versions.

- Private collaborators — you invite someone to edit one specific document. They see that document, not your workspace.
- Share the tree — when publishing or sharing a document you can include its whole branch, to hand over an entire manual rather than a single page.
- Public link — publish and anyone with the link can read it, no account needed. Unpublishing cuts access instantly.
- The public link is stable: republishing the same document always returns the same address, so the link you sent three months ago is still the right one.

## Generating and editing with AI · Hobby

What sets this apart from pasting your request into a generic chat is context: here the model works on your material — your documents, your notes, your recaps, your files — and not on whatever you manage to paste into a prompt.

- **A full document from context** — You say what you want and from which sources, and it writes the whole document: brief, outline, report, proposal. With references to what it used.
- **Your own templates** — The same summary templates you define for recaps can be used when generating a document, so the output always comes out in your format.
- **Create from an image · Pro** — A photo of a whiteboard or a sketch on paper becomes a structured document — or a canvas, with the boxes and arrows reconstructed.
- **Rewrite a selection** — Select a paragraph and ask to shorten it, formalize it, translate it or explain it better. It changes what you selected and nothing else.

_Generating means choosing what context goes in — child documents, recaps, canvases, files — how long and what kind. It writes from your material, not from what the model remembers._

## Canvases

An infinite Miro-style space for what does not fit in text: maps, architectures, relationship diagrams, sketches.

- Shapes, connectors, free text, sticky notes and images.
- Freehand drawing with Apple Pencil on iPad, with stroke pressure.
- A canvas is shared and published exactly like a document.
- On mobile they open in a dedicated viewer, with zoom and pan.

## Files

Everything you upload — to a document, to the daily note or through the WhatsApp chat — lands in Files.

- Inline preview of PDF, Word and Excel, without downloading anything or leaving the app.
- Text extraction: the content of a PDF or a Word file enters search and can be used as context for generation.
- Rename, move, crop images and delete, with your storage usage in view.
- On mobile, any file opens in its own viewer.

_Everything you uploaded and everything the product generated, in one grid, with the same search inside them._

## Search: one box for everything

Search is hybrid: it combines literal matching with semantic similarity. You find what you wrote with the exact words, and also what you wrote with different ones.

It sweeps documents, daily notes, recaps, tasks and the text extracted from your files, all together, and returns each result with its context and its origin.

## Integrations

- **MCP** — External AI clients (Claude, ChatGPT and others) can query and operate your workspace over the MCP protocol, only with the token you generate.

Over MCP, an external AI client can list your tree, read a document, create a new one, update it, move it, search across all of them and link it to Atlas entities. See MCP.
