What is a shared brain?

A git-backed vault of markdown files that LLM agents keep up for a whole team, holding tasks, ideas, competitive intelligence, specs, objectives, people profiles, and agent memory in one linked graph.

How does it work?

Agents do the writing and upkeep. People curate sources, make decisions, and ask questions. The vault keeps per-user memory the agent writes from observation, and team memory of decisions that lasts across users and sessions. A new addition gets checked for contradictions, cross-referenced, and pushed through to related tasks and ideas.

Where does the human gate sit?

Automated health checks run across seven categories, including contradictions, stale claims, and orphan pages. Items that can be fixed automatically get fixed. Issues that need judgment go to a human.

What does it change for the team?

Knowledge compounds for everyone, not one person. A product manager’s competitive research surfaces when a salesperson asks for deal context. An engineer’s documented constraint shows up during spec writing without a Slack message. A wrap up at the end of a session writes decisions and open items back to the vault, so nothing is lost in chat history.

Which essays cover it?

• The Shared Brain

@misc{couch2026whatisasharedbrain,
  author = {Couch, Chris},
  title = {What is a shared brain?},
  year = {2026},
  howpublished = {Lead to Cash},
  url = {https://chriscouch.ai/what-is/shared-brain/}
}