Brain · The Brain Platform
Bring a shared knowledge layer to compatible AI tools, so the next conversation can begin with the context your team has chosen to connect.
Claude: What decision did we make about the launch date?
The fictional launch note records a move to 14 November so the support handoff can finish first. In this illustration, Claude consults the same dated source. Real model wording and available sources can differ; check the note and its owner before acting on the schedule.
Source excerpt · Fictional launch decisionDecision: move launch to 14 November, after the support handoff. This dated note is the maintained source for launch timing.
AI memory is retained context that an AI assistant can consult beyond one conversation. In Brain, your team connects a shared knowledge layer for compatible clients to consult, with sources you can review.
AI memory is the context an assistant can use beyond a single prompt. When a project brief lives in one chat, a decision in another and a policy in a shared drive, people spend time reconstructing the same background. A teammate may have the files but still lack the explanation that makes them useful.
Brain gives connected knowledge a place that compatible AI clients can ask. The benefit is a consistent source to consult across work, with citations that let people inspect where an answer came from. It complements the memory features inside individual assistants; those features have their own scope and settings.
Connect. Ask. Govern.
Choose accepted project decisions, policy references and working assumptions that the team owns. Keep an owner and review date beside each source so shared memory does not turn an obsolete conversation into current guidance.
Ask which decision applies to the next task, then inspect its source. Use another compatible client to consult the same maintained collection rather than assuming every conversation transcript was automatically saved.
Review workspace membership and source scope before sharing the collection. Compare a permitted teammate with a restricted one and keep confidential project material outside the ordinary team-memory path.
See the idea in action
Fictional examples. These interactions do not query your Brain or test real permissions.
The fictional launch note records a move to 14 November so the support handoff can finish first. In this illustration, Claude consults the same dated source. Real model wording and available sources can differ; check the note and its owner before acting on the schedule.
Decision: move launch to 14 November, after the support handoff. This dated note is the maintained source for launch timing.
The fictional launch note records a move to 14 November so the support handoff can finish first. In this illustration, Cursor consults the same dated source. Real model wording and available sources can differ; check the note and its owner before acting on the schedule.
Decision: move launch to 14 November, after the support handoff. This dated note is the maintained source for launch timing.
The fictional launch note records a move to 14 November so the support handoff can finish first. In this illustration, an authorized teammate consults the same dated source. Real model wording and available sources can differ; check the note and its owner before acting on the schedule.
Decision: move launch to 14 November, after the support handoff. This dated note is the maintained source for launch timing.
The fictional decision records a move to 14 November, after the support handoff. In the limited-access example, the teammate is not assigned this source, so both this answer and its source excerpt are withheld. This is an illustration of a denied state, not a live permission check.
Decision: move launch to 14 November, after the support handoff. This dated note is the maintained source for launch timing.
Open a question, then inspect its source excerpts.
Brain is a knowledge layer reached through MCP, the Model Context Protocol. Claude, Cursor and Codex are examples of compatible clients in the existing product. Client support and setup differ, so check the current connection instructions rather than assuming every assistant has the same capabilities.
The current public setup describes Google Drive and Notion as connected sources, alongside documents you choose to add. Start with a focused collection and inspect the actual connection screen for availability. A tool appearing in a roadmap or illustration does not mean its connector is ready for your account.
Connect the documents your workspace needs.
Bring approved pages into a shared knowledge layer.
Add a focused set of knowledge you own.
Keep access intentional
A retained decision can be useful to several teammates without being available to everyone. Choose the workspace and source membership deliberately, and inspect a restricted request using approved examples. Brain’s connected collection is a maintained reference layer; a source owner should correct outdated guidance and decide when a private decision can become shared knowledge.
Fictional metadata only. No document content is shown, and this is not a record from your account or proof of live enforcement.
Illustrative history indicator: maintained decision consulted; full chat history not reloaded in this example. Claude, Cursor and the teammate cards point to the same fictional source rather than reconstructing a separate project history. This is a presentation example, not a measured reduction in input tokens or an API trace. Actual client context, retrieval and costs depend on the configured workflow and must be evaluated separately.
Compare current plans and limits| Approach | What to consider |
|---|---|
| Memory inside an assistant | Useful for that assistant’s continuity and personalization. Its availability, sharing and controls depend on the provider and plan. |
| Project instructions and files | Useful for work scoped to a project. Someone still needs to maintain the files and decide who can use them. |
| Brain as a shared knowledge layer | Gives compatible clients a connected source to ask and inspect. The team still owns source quality, access settings and review. |
AI memory is information an assistant can carry into later work rather than receiving it again in every prompt. It can include preferences, prior decisions or connected knowledge. The useful question is which information is retained, where it lives, who can use it and how an incorrect detail gets corrected.
Claude offers memory with scope and controls that depend on its features, plan and settings. Its official documentation describes separate project memory and organization controls. Brain adds connected source material for compatible clients to consult. Compare the knowledge you want shared and maintained, rather than assuming an assistant begins every conversation without context.
ChatGPT supports personal memory and shared projects with their own settings and boundaries. Shared projects can bring collaborators into the same files, instructions and conversation history. Brain provides a separate connected knowledge layer for compatible clients. The distinction is how your team maintains source material across tools, rather than whether ChatGPT can share context.
A team can maintain shared knowledge instead of leaving useful context in separate private chats. Decide which documents belong in the shared workspace, assign owners and test the intended access. Sharing knowledge does not mean every member should see every document, and a useful answer should still point back to its sources.
Brain exposes knowledge through MCP, so the client needs a compatible connection path. The existing product describes Claude, Cursor and Codex as examples. Capabilities and setup vary by client and account. Consult the current connection instructions and test a simple cited answer before making that client part of your team’s workflow.
Projects can remain useful for focused instructions, files and conversations inside Claude. Brain adds a separate connected knowledge layer that compatible clients can consult. Use a project for the work you want scoped there and shared knowledge for material that should be maintained across the team. Avoid keeping conflicting copies of the same policy.
Try the public demo, then choose a small, well-maintained collection for your first shared knowledge workflow.
Fictional launch decision