What Is a Brain?

A brain is an area of life. It is not a project, and it is not a model.

Work is a brain. Family is a brain. You might add one for a property, a client, or a trip. You would not make a brain for every repository, every chore, or every document. Those are pieces of work that live inside a brain. The partition is the point. If everything is its own brain, you have rebuilt the mess of separate chats with extra steps.

Each brain has its own agents, its own rules, its own memory, and its own approval queue. An agent you trust with work is not automatically trusted with family. There is no cross-brain search. The wall is in the data, not in a hope that every session will behave.

The rules are different because the stakes are different. Work can say that nothing is purchased and nothing is published without your yes. Family can say that plans stay inside that brain. Memory follows the same line. A decision about a release does not belong next to a note about Saturday, and a token issued in one brain cannot read the other.

Creating a brain is a short form: a name, and a sentence about what belongs there. After that, you connect the agents that should work inside it. Buddy might be a push agent on the work brain, woken by a signed webhook. Claude might be a pull agent, checking an inbox when you ask. Shadow might be a relay agent, because the chat he lives in has no connector, so you carry a signed link by hand.

iBrains is neutral about which AIs you use. It does not sell intelligence, and it does not sell brains. There is no marketplace of brains, and there will not be one. A plan, when accounts are open, is a limit on how many brains you can run. It is not a way to buy or hire a brain.

The trust story is the partition. You decide what each part of your setup knows. An agent can be capable and still be kept out of the room where it does not belong.