One brain for everything is one brain for nothing.
If you run a consulting practice, you probably hold three or four conversations a day that have nothing in common. A client retainer. A marketing experiment. A new framework you are sketching. A founder you are mentoring on the side. Drop all of that into one AI knowledge base and search starts answering the wrong question. The brain pulls a client quote into a marketing idea. The wrong project gets cited at the wrong meeting. Worse, you stop trusting the answers, because you cannot tell where the context came from. The fix is not less capture. The fix is brains with edges.
How multiple brains work
Spin up a brain per stream
Create a brain for each meaningful stream of work. Personal life. One per client. One for your content engine. One for the side project. Each brain is its own knowledge graph with its own entities, its own documents, and its own history. They do not bleed into each other.
Capture straight into the right one
Every upload path lets you pick a brain. The web app, the API, the MCP connector, and Zapier all accept a brainId. Your default brain catches anything you do not route. Make a habit of routing the client work, and the marketing material stops appearing in client briefs.
Search across all, or narrow to one
By default, every question spans every brain. That is the right default when you are exploring or connecting ideas across streams. When you need the answer to come from one brain only, name it: 'search just my Marketing brain.' Both work from the same connection, no switching.
Who multiple brains are for
Consultants juggling clients
One brain per client. Every transcript, every doc, every framework lands in the right space. Client A's pricing experiment never shows up in Client B's briefing.
Coaches with public + private work
A Personal brain for journals and decisions. A Coaching brain for client material. The marketing brain for posts you draft. Ask each one its own questions.
Builders with parallel projects
A brain per product, a brain per book, a brain per research thread. Voice memos, articles, and meeting notes go to the brain they belong to. Patterns surface across all of them when you want.
What you walk away with
Route to the right brain from where you already work
From Claude, via the MCP connector
The connector exposes list_brains. Say "search my Marketing brain" or "save this to my Client A brain". Claude routes correctly.
From Zapier, with a brain id
Every Brain Graph upload action accepts a brainId. Point each Zap at the brain it should fill. No more cross-pollution.
Frequently asked questions
How is this different from having two accounts?
Can two brains share a document or entity?
What does the default 'search across everything' look like?
Can I limit a brain to a specific person on my team?
Does this work from Claude and Zapier?
Will my Free plan limit apply to each brain separately or in total?
Can I rename or delete a brain later?
One account. Many brains. Every kind of work.
Start free with 40 documents. Split them across as many brains as you want.