> ## Documentation Index > Fetch the complete documentation index at: https://docs.monnet.ai/llms.txt > Use this file to discover all available pages before exploring further. # Company brain > Your team's wiki — what you know, kept true over time, maintained by Monnet from closed motions and the sources you upload. The **Brain** is your workspace's wiki. Motions record what happened; the brain records what your team ended up believing. Monnet maintains it, and reads it whenever it facilitates — so the more the brain knows, the less you have to re-explain. Open it from **Brain** in the sidebar. ## Two tabs **Wiki** The pages, in a tree of folders. Pages and folders sit side by side under a parent and you can reorder them by drag and drop. A page is a document like any other in Monnet: it has comments, version history, and its own one-to-one chat with Monnet. You can edit any page by hand — Monnet writing a page first doesn't make it Monnet's. **Sources** The raw material the brain is built from: files you upload, and the closed motions eligible for the next ingest, split into what's already been digested and what's queued. Sources are the record. They're read, never edited. When a page looks wrong, open the source behind it to check — then correct the **page**. ## The pages your workspace depends on Two kinds of page are loaded automatically and can't be created, renamed, moved, or deleted by anyone. Only their contents are open. * **The page named after your workspace** holds your **Policies**: how the team works, what it prioritizes, who decides what. Monnet reads it on every engagement. This is where the workspace-wide brief lives. * **Each page under `Team/`** is one person: how they work, what they own, how to engage them, what's on their mind. > [!NOTE] > > A profile page is written for the whole workspace to read, and its subject reads it too. Monnet builds it from the team's own record — documents, notes, closed motions — never from someone's private conversation with it. ## Choose where Monnet messages you To guide where Monnet sends you private messages it starts, open **Brain → Wiki → Team**, find your page, and write your preference in plain language. For example: > I use the Monnet web app throughout the day. Please send me messages there instead of Slack. Monnet considers your profile alongside the motion and the message when choosing between channels that can reach you: the web app, Slack, Microsoft Teams, or email. This applies to messages such as plan-step assignments, check-ins, and follow-ups. You don't need to disconnect Slack to keep using it for replies. This is guidance, not a guaranteed mute. If your page has no channel preference, or Monnet can't use it to make a choice, it falls back to its usual channel order, which favors Slack when available. An explicit request to send a message on a particular channel can also take precedence. Replies still arrive where you contacted Monnet, and [@mention pings](/product/discussion#mentions-reach-people-where-they-are) still use their own channel order. There is no separate notification or quiet-hours setting. ## What gets ingested, and when Every night Monnet reads what has landed since its last run and folds it into the wiki. It only runs when there's something new. It reads: * **Sources you uploaded** that haven't been digested yet. * **Closed motions shared workspace-wide.** A restricted motion is never ingested — [sharing it with the workspace](/product/motion#sharing) is what makes it eligible. A workspace's first run digests its whole history, so the brain doesn't start blind to everything decided before you turned it on. Control it from **Settings → Scheduled** under **Workspace**: switch the nightly run on or off, and customize its instructions. Switching it off stops the brain being maintained — nothing new is digested, and nothing already written is removed. You can still run it by hand from the Brain. ## How pages are written There's no fixed taxonomy, and the rules aren't ours — they live in your workspace's own wiki schema, which you can read and rewrite. The defaults: * **A page compresses a subject**, pulling together facts scattered across many motions and files. Never one page per item — that's a filing cabinet, and you already have one. * **Most new material belongs on a page that already exists.** Monnet reads the wiki before deciding anything is new. * **When new material contradicts a page, the page is corrected at the source** — not annotated with a correction underneath. * **Nothing is marked settled unless the material says so.** Something raised but unresolved is worth a page saying exactly that. Pages link to each other, and the app computes the reverse direction — so a link written once is visible from both ends. ## Who can see it The brain is workspace-wide. Everyone in the workspace can read it, and on a page shared with the workspace, everyone who can read it can edit it. Guests — people invited to a single motion from outside your workspace — never see the brain. Monnet won't read it or write to it while acting on their behalf. ## Start with company context During [web onboarding](/product/onboarding#seed-the-brain), share public website or product documentation links and upload files that explain your product. Monnet builds a basic company brain from those sources before reviewing your first idea. Open **Brain** to review what it learned, correct incomplete or incorrect pages, and add more documents in **Sources**. Include current priorities, your roadmap, and user research so reviews reflect what matters to your team. ## Start with repository context [Onboarding from Claude Code or Codex](/product/onboarding#start-from-your-coding-agent) uses supported context from your work folder to fill in the existing workspace and personal profile pages. It also creates one page for each product you own, giving Monnet the context it needs to assess incoming product ideas. The briefing appears in **Sources**, and onboarding requests an immediate wiki update. Pages can take a few minutes to appear. If the update cannot start, the source stays available for a manual update from Brain or the next enabled nightly run. ## Reading it from your coding agent The [MCP server](/integrations/mcp-server) exposes the whole brain: browse the tree, read and search pages, create and edit them, and manage sources — from Claude Code, Claude Desktop, Cursor, or Codex. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.