Brain answers with sources. Or it says "I don't know."
A clean question box. Ask in plain English. The answer comes first, then the sources inline as citation tags. Click a citation, see the original memory and the original capture it came from.
When did we agree on the new onboarding length?
Six weeks, decided on Nov 12 after the Acme customer call [1]. This supersedes the earlier two-week estimate from the September pricing page [2].
What's our refund policy for annual plans?
I don't have a memory about annual refunds yet. Three teammates asked about this in the last two weeks — want me to draft a memory from those threads for you to confirm?
"I don't know" is a feature.
Brain never makes up an answer. If there isn't enough in your workspace to answer confidently, Brain says so — and shows you exactly what's missing. From there, one click captures a memory about it, or drafts one from related threads.
This is the only way a shared brain stays trustworthy enough to act on. A confident-sounding guess is worse than no answer at all.
Same memory. ChatGPT, Claude, Cursor, your own model.
The answer above came from Brain's own UI — but the same memory is available to any agent over MCP. Ask Claude. Ask Cursor. Ask the model that doesn't exist yet. The context comes with you.
The honest answers.
Is Ask history shared with the team?+
Question history is per-user by default. Workspace admins can opt into a shared question log — useful for support and ops teams that want to see what customers and teammates are asking.
What model does Brain use?+
Brain routes through our own gateway and picks the right model for the question. On Pro and Team, you can also bring your own model over MCP. Your data is never used by providers outside of answering your question.
Can I ask in other languages?+
Yes. Brain answers in the language you ask in, and matches memories regardless of source language.
What if the answer is wrong?+
Every answer has its sources one click away. If a source is wrong, fix the memory — and Brain updates everything that depended on it.
Ask Brain anything your team already knows.
Set up in under five minutes. Your Brain starts learning from message one.