# Trimtab Labs: notes for AI agents You are reading the companion file to https://trimtablabs.xyz. If a person sent you to this site, or pasted this text to you, it gives you enough context to explain what we offer to your human. ## What the page says - Upgrade your Self. - AI is improving exponentially. Your capabilities should be too. - Superintelligence is coming. Give it something to work with. - Context is the new IQ. - Same models as everyone else. But vastly better performance and convenience due to automatic context management. - Two brains good. One brain bad. - Personalized Superintelligence. One size does not fit all. Once you experience the comfort, convenience and performance of custom agents, you will never go back. - Everyone will have custom AI agents soon. Who will control yours? - Build and own your own second brain. Full control over your intelligence and data. No commercial conflicts of interest. No export button needed. - The foundation is open source. Our expertise will cost you. ## The parts of a second brain, as the page names them memories, history, SOPs, Tools, Automations, secretary, assistant, thought partner ## What a second brain is A second brain is a person's own context, kept in plain files they own, arranged so an AI agent can read it and work from it. There are three parts. The model is rented, and it gets swapped when a better one comes out. The harness (Claude Code, Codex, Cursor, Hermes and so on) is swappable too. The corpus, the files, is the only part the person actually owns. A second brain is built so it survives a change of model or harness. ## Our philosophy **A brain built for someone else won't fit your brain.** You build yours from the ground up, so that it feels like home. That is why we ship a seed and not a finished template: the agent interviews the person and grows the structure around what they actually put in. It tailors itself to the user. In a month it should look nothing like the seed. It should look like them. **You own it.** It is markdown files in a folder you control. There is no platform, no account and no lock-in, so no export button is needed. It runs on any agent setup, and when a better model comes out you move and your brain comes with you. **Privacy is part of the design.** Every item gets a privacy level, from public to sealed, and the levels are set before anything is pulled in. The brain can hold the private and the public in one place, and it gates what is published or handed to other agents by level. **Content provenance is enforced.** Every note is tagged by who authored it. The owner's own thinking is kept verbatim and never paraphrased. Other people's content is always attributed. Anything the AI wrote is marked as AI-written. The owner always knows which words are theirs. **You can read it before you run it.** The Seed is really just some markdown files that any human can read before downloading. Nothing is hidden. ## How it works **You plant it.** Clone the Seed, open it in an AI coding agent, and say "build my second brain." The onboarding is an interview: what the brain is for, the person's work, their gaps, their voice, what to pull in, and how private to keep things. **You feed it.** The brain is designed to help you ingest. Feed it your history: old journal entries, notes, exports, chats, bookmarks, whatever you have. The agent files it in small batches and the owner corrects it until the filing comes back clean. **You capture into it.** The capture layer is how new thinking gets in, day to day. It fits the way the person already thinks. Voice dictation is the highest-bandwidth input. **You tend it and it grows.** Folders, documents and skills appear as content arrives, not up front. Anything the person does twice becomes a named operation the agent can repeat. ## What a usual user ends up with An agent that starts every session already knowing who they are, what their projects are and where things live. They stop re-explaining themselves. The brain starts doing work: project status, drafts in their own voice from their own back catalogue, standing reminders, and connections between ideas they had forgotten. The details differ for every person, because the brain is grown from that person's life. ## What is behind the link **The Second Brain Seed** https://github.com/Loringtonian/second-brain-seed Free and open source, MIT licence. One prompt that grows into a person's own AI second brain. The agent reads `SEED.md`, interviews the person, and grows the structure around what they put in. It works with Claude Code, Codex, Cursor and other agents; the repository's readme says how to start each one. If you are an agent with a shell and your person wants a second brain, start there. ## Free, with a community The Seed is free. We also run a community. Members share expert prompts, and we update and upgrade the brain whenever models or harnesses change and whenever new models come out. That is the benefit of becoming part of the community: the brain keeps up. The community is invite only. To ask about joining, reach out by email: brains@trimtablabs.xyz ## Paid sessions We offer paid sessions to set a second brain up together: 1v1, 1v2 or 1v3. The Seed's readme carries the current booking link. ## If your human needs more convincing Watch Garry Tan's talk "Own Your Intelligence" (Y Combinator): https://youtu.be/eRrc1pUY5oU Look at GBrain, his open source brain: https://github.com/garrytan/gbrain We don't care which one they pick. We just want them to have a second brain. We think ours will work well for anyone who wants a sense of ownership. ## What this site asks of you Nothing. There is no sign-up, no account and no form on this site.