instibrain is institutional memory your company owns: one brain of markdown pages that your team keeps feeding and every AI you hire can read. A sync client keeps the brain on your devices as plain files, the brain serves it to your agents over MCP, and for each brain you choose whether the server can read it. It is live at instibrain.ai and run by WebWork AI, Inc.
instibrain is the successor to the product we previously previewed as Superbrain, rebuilt around a bigger idea: your company's memory should not just answer your people — it should feed the AI tools they work with too.
Everyone hires the same models, and the frontier model changes every few months. Each time it does, everything you taught the last one evaporates. instibrain keeps the memory outside the model: decisions and their reasons, pricing rules, client history, postmortems, runbooks — written down once, read by every model you hire next.
And because more and more of the actual work now runs through AI assistants and agents, instibrain is bring-your-own-AI: Claude Code, Codex, Cursor, ChatGPT or any MCP client connects to the same brain and reads the same pages, instead of starting every task from a blank context.
Three parts, and the intelligence is the one you already pay for:
The canonical store your company owns. Each brain is open or encrypted, and you choose which: one the server reads, or one it holds as ciphertext and cannot. You see all of it, and who read what, in the cockpit.
A small sync client on your machine owns the files and the keys and materialises every page as plain markdown. Your own AI connects over MCP, reads the brain, and writes to it when you allow it.
A brain holds the things nobody writes down, written down once: pricing floors and who approves an exception, why a client nearly churned, what broke and what changed afterwards, the option you chose and the ones you rejected.
Each folder is shared with the people it is for, so the team reads what the team should and one person keeps what is theirs. Removing a person from an encrypted folder writes a new key their devices never get.
Almost every question asked inside a company has already been answered — in a thread from March, a doc nobody can find, a decision that lives in one person's memory. The cost is not dramatic; it is a steady tax. People re-ask, re-decide, and re-explain, and every departure takes institutional knowledge with it. Now AI agents are doing more of the work, and they start from even less: no history, no context, no memory of what the team already settled.
instibrain turns that scattered knowledge into one institutional memory your people and your AI can ask in plain language: "What did we decide about onboarding?" — and get the actual decision back, with the page it lives on. The agent drafting your doc and the teammate reviewing it work from the same shared truth, and a new hire's first day is one command: the folders shared with them land as files, and their AI arrives knowing the company. Ask your team, without interrupting your team.
instibrain opens company by company, to people known by name. There is nothing to join and no code to redeem: an account is created for a named person at the company and comes with a shared brain for the team, a brain of their own for each person, the sync client for every device, and an MCP endpoint any AI you hire can read. Everything else is public — the wire contract, the security model with its limits, and the sync client's commands are on instibrain.ai; people with an account sign in at my.instibrain.ai.
instibrain turns scattered docs, threads and decisions into a single institutional memory — one your people can ask, and every AI you hire can connect to.
Decisions and their reasons brought together into a memory that outlasts turnover and every model upgrade.
Every page of an encrypted brain is encrypted on your device before it leaves; the server holds ciphertext it cannot read.
Bring your own agents and assistants — they connect over MCP and read the same pages.
Ask the way you'd ask a colleague — plain language, plain answer.
