The Third Brain · Coming soon
Put the notes you never open again to work, without learning to code
This course sets up a small team of Claude Code agents on a copy of your notes. One files new material, one suggests links, one drafts from a set of notes. They write only to a review folder, and nothing reaches your notes until you approve it. You type requests in plain English; no programming.
The course: The Third Brain: a Claude Code agent swarm on your notes, for non-engineers
$39 $79 early-bird price
This course is being built now. Join the early-access list to reserve the early-bird price and shape what it covers.
Joining the list is free. You only pay if you buy at launch, at the early-bird price.
If this sounds like your week
- Your vault keeps growing, and most of what goes in is never opened again.
- You tried atomic notes, tags and folder systems, and the upkeep became the work.
- When you sit down to write, you start from a blank page even though the thinking is already in your notes.
- You have seen Obsidian and Claude Code setups online, then stalled at the first terminal command.
What you do now, and where it stops working
Today you capture, promise yourself a clean-up weekend, and paste the odd note into a chat window when you need it. A second brain stores your thinking, but it waits for you to act on it, and you rarely have the hour that takes.
What changes after the course
- New material lands in an inbox and gets filed in the right folder, with the move waiting for your yes.
- You get a short list of suggested links between notes each week and accept the ones that make sense.
- You ask for a first draft on a topic and get one built from your own notes, with the source notes listed.
- You can read every change an agent proposes before it happens, and undo any you accepted.
How it works: The approve-first swarm (the third brain)
- 1Back up the vault and give the agents a working copy, so your originals stay untouched while you learn.
- 2Write one plain-English instruction file per agent, each with a single job: capture, linking, summary or drafting.
- 3Agents write only to a review folder; your notes are read, never edited.
- 4Run a weekly review: accept, edit or discard what the agents left, and only accepted changes move into your notes.
What you get
- Starter vault with the folder structure the agents expect
- Agent definitions and instruction files for each agent
- Approval checklist and weekly review routine
- Troubleshooting guide for common setup problems
- Video walkthrough of the full swarm on a real set of notes
This is for you if
- You keep notes in Obsidian or plain text files and want them to produce something.
- You use Claude or ChatGPT already and are willing to try Claude Code with step-by-step help.
- You want AI working on your notes only where you can see and approve every change.
Skip it if
- Your notes live only in Notion, Evernote or Apple Notes and you do not want to export them to files.
- You want an agent to rewrite your notes for you with no review step.
- You already build Claude Code agents; the course will move too slowly for you.
Before you join
- Will it mess up or delete my notes?
- The first module has you back up the vault and give the agents a copy. Agents write only to a review folder and never edit your notes directly. A change reaches your notes only when you accept it. Jay West runs his own work on this pattern: every agent change is reviewed before it ships.
- I cannot code. Can I still do this?
- Yes. The course is written for people who have never used a terminal. You type requests in plain English, every command you need is shown to copy, and the agents are defined in plain-language instruction files, not code.
- Are my notes private?
- Your notes stay as files on your own computer. The parts an agent reads are sent to Anthropic to be processed, as in any Claude conversation. The course shows you how to keep private folders out of the copy the agents can see.
- What does Claude cost on top of the course?
- Claude Code runs on a paid Claude account, and that cost is separate from the course. The course does not sell or include a Claude plan, and it starts with small runs so you see the usage before you rely on it.
- My notes are a mess. Is that a problem?
- No. A messy vault is the starting point the capture and linking agents are built for. You do not tidy anything before you begin.
Rather have it built for you? Describe the project and the same work can be done for your team.