Load a Brain
Join the waitlist

Free mini-brain · brains launch Tuesday 1 December 2026

Give your AI agent a brain it can cite.

A brain is a folder of sourced, linked notes on one topic, with a small system that makes your AI agent answer from those notes first. Join the waitlist and get a free one now: the Starter, 42 basic notes from six of our brains.

From Starter to full brain

Start with the basics. Slide to see the rest.

The Starter is a real slice of our brains: the same notes, with the same sources. Slide to load the full brains around it.

ringed dots = notes in your free Starter · squares = skills

Your free Starter

What your agent can read

42sourced notes
6brains
0mapped areas
0skills
Lite OSsystem: Lite OS searches and quotes; Brain OS adds the skills

What a brain is

Notes your agent can trust, and a system to use them.

01

Sourced notes

Every note lists its sources with dates, and how sure it is. Your agent quotes them back to you.

02

Linked into a map

Notes link to each other, so you and your agent find the right one fast. Open the graph in Obsidian and see the whole topic.

03

An engine with skills

Brain OS tells Claude Code (or any agent that reads files) to search the notes first and the web only after, and gives it skills: ready jobs like "prepare a sales call". The free Starter runs on Lite OS: search and quote only.

The free mini-brain

The Starter: the basics from six brains.

Seven basic notes from each of six brains, the ones the rest of each brain builds on. Each brain also has a door note that shows what its full version covers.

It runs on Lite OS: your agent searches these notes and quotes them. That is all. The full brains run Brain OS.

Free. One zip, for Obsidian and Claude Code.

  • Sales: opening a conversation, running it, closing7 notes
  • Marketing: one landing page, a small first test, measuring it7 notes
  • Web: a fast site that search and AI can find7 notes
  • Brand & Design: what makes a brand recognisable and readable7 notes
  • Finance: pricing and cash for a small business7 notes
  • Agentic: using agents safely and judging AI claims7 notes

A real note from the mini-brain

This is what your agent reads.

Short, dated, with its sources and a trust level. Every claim points to its source.

notes/Web · confidence: high · sources dated 2024–2025

Core Web Vitals

The three user-experience metrics Google uses for loading, responsiveness and visual stability, with their good thresholds at the 75th percentile of page loads: LCP 2.5 s or less, INP 200 ms or less, CLS 0.1 or less.

Sources: Web Vitals, web.dev (2024) · Understanding Core Web Vitals and Google search results, Google Search Central (2025)

How it works

Three steps, about two minutes.

01

Download the zip

The link comes by email. One folder with the notes and the engine.

02

Open it in Obsidian

Free, on Mac, Windows and Linux. The graph is there.

03

Start your agent inside it

Run claude in the folder and ask. It answers from the notes, with sources.

Privacy

Your files stay on your computer.

The brain

A brain is plain files on your computer. It sends nothing anywhere, and we never see what you add to it.

Your AI provider

When you use an AI agent, the AI provider (for example Anthropic for Claude Code) sees the notes the agent reads to answer you, under that provider's own terms.

Notes on laws and money quote their sources and dates. They are not legal or financial advice.