Living Brain

The problem

What your team knows is spread across chats, agent sessions and heads, and the wiki goes stale.

What it does

Living Brain is a team wiki that writes and maintains itself from your conversations.

How it works

1. Team conversations and agent sessions become Markdown pages for people, projects, decisions and customers, every fact linked to its source message
2. Nightly passes merge duplicates, surface contradictions and refresh stale facts
3. It reads only with the asker’s own access, over MCP and in the CLI
4. Team chat in Slack and Discord is planned

TodayLive in early access: the web app at app.livingbrain.wiki (open sign-up), the MCP endpoint, the Mac app and the livingbrain CLI syncing Claude Code, Codex and Colonizer sessions. Team chat, other agent harnesses and paid plans are planned.

The film

Aims

Autonomous operation

The wiki is the thing that should keep itself: conversations become pages with their sources linked, and a nightly pass merges duplicates, surfaces contradictions and refreshes stale facts, with no one assigned to tend it. People ask questions and correct it; they do not write it.

Intelligence

Models read team conversations and write the pages, citing the source message for every fact; the nightly passes reconcile what changed, and a learning layer models how each person works. Teams bring their own LLM, LiteLLM included. Reading is always done with the asker’s own access.

Growth

Team-led, inside the tools people already use: coding agents over MCP, the terminal, and team chat (Slack, Discord). A free self-hosted tier for small teams first. Growth is workspaces with a wiki that is being kept, counted from the first one that is not ours.