Everyone uses the same models. The difference is what yours knows.
Your model is everyone's. What it knows about your company is yours alone, and it knows only what your team wrote down.
Anyone can use the models on the left. Only your AI reads the pages on the right.
Your AI stops answering generically about your company.
A recurring task starts from the current page, not from an old chat. Ask about a client, a deadline, or a rule, and the AI you use reads the page first.
- It reads firstThe brain's first rule to every AI: read the space's
CLAUDE.mdbefore anything - It can name the pageAn answer that says where it came from is one you can check
- You explain the company onceNot again in every session, to every AI
What is open on the Lisbon office, and who owns it?
search_brain Lisbon · read_node roadmap.mdread
Who owns it: Maren. The roadmap's owner table (roadmap.md) lists the
Office item as Maren's, due in November. It's goal #2 for Q4 2026, “Open the Lisbon
office.” … What's open: the Lisbon lease renewal. An inbox item from 2026-10-25
says “call Tobias about the Lisbon lease renewal before Friday.” Nothing shows
it was done.
Fix the page once, and every AI reads the fix.
A procedure has one page. Edit it in the browser or in your editor, and every AI of everyone who reads the space gets the new text on its next read. Nobody keeps a copy of their own.
- One page, edited in placeIn the cockpit, or as a file on your computer
- Every AI, next readClaude Code, Codex, and Claude on a phone read the same version
- The old text is keptA version, with its writer; a bad edit is never the last word
Change the model. Keep what it knew.
Switching models used to cost everything you taught one. The next model reads the same pages on its first day, from the same address.
Read what your AI reads, and correct it.
The markdown your AI reads is on your computers and in the cockpit. Open it, see who wrote each block, change a line, and keep the folder with or without us.
- Files on your computers
~/instibrain/spaces/, kept in step by the sync client - Every writer namedA person, a computer, an AI, or an agent, on each version
- No model trained on itYour pages train no model and are never sold
Choose who reads each space, and what our server can read.
A space gives one body of knowledge one audience. A sealed space adds protection from us: encrypted on your computers, ciphertext here.
- One audience per spaceShared whole; what only some people should read gets a space of its own
- Sealed: 0 bytes we can readEncrypted before it leaves your computer, the path included
- The honest sentenceThe host cannot read a sealed space. Your computer and the model see what they read
A wiki, a search, and the vendor's memory each do a job.
Each is a product someone else makes well. Four questions tell them apart from a brain.
| The question to ask any of them | A wiki | A search over your tools | The model vendor's memory | A brain |
|---|---|---|---|---|
| Who can read the plaintext? | The vendor, and everyone with access | The vendor's index | The model vendor | Open: us and its readers' AIs. Sealed: its admitted computers only |
| What do you keep if the vendor vanishes? | An export, if one is offered | Nothing; the index was theirs | Nothing | The folder of markdown already on your computers |
| Which model is it tied to? | None, and none reads it | The vendor's | That vendor's | Any AI you connect, this year's or next's |
| Who corrects a wrong line? | Whoever notices, if they may | Nobody; it is an index | Not visible to you | Anyone who reads the page, with every version kept |
What your team writes down today, every AI you use reads tomorrow.
Request access for your company, read the use cases, team by team, or see who builds instibrain.