Build a second brain in Obsidian your AI can actually read.

You can't connect the dots looking forward — only looking back. But you can only look back if you wrote it down. Most people can't connect their dots because they don't have the dots. And an AI knows you exactly as well as you've written yourself down.

This is the setup I actually run. Plain folders, plain text files, no subscription, about twenty minutes to build. Everything below is the real thing, not a demo.

1. Install Obsidian

It's free, and your notes stay on your machine as ordinary Markdown files — no lock-in, no export needed, readable in ten years. Get it at obsidian.md, create one vault, name it anything.

2. Four folders, and no more

• 10-Projects — things with an end date
• 20-Areas — ongoing, no end date
• 30-Resources — reference material
• 40-Archive — finished, never deleted

Resist adding more. Depth is where second brains go to die — every extra layer is another decision you have to make before you can write anything down.

3. Every note gets a header

A few lines of YAML at the top of each note — id, type, tags, created, updated — then a title, one sentence saying the single idea, and the body. Link related notes to each other as you go.

4. Write the rules file — the part everyone skips

A second brain without rules becomes a junk drawer in about three weeks. Four rules in a CLAUDE.md at the root are the whole system:

• Frontmatter always. Every note carries the header. No exceptions.
• Links always. Every note connects to at least one neighbour. An orphan note is a lost note.
• Never strip context. Only add or update — never delete. Bump the updated date every edit.
• Mark what expires. Every fact is timeless, dated “(as of 2026-08)”, or a pointer to a live source. This is the one that keeps it true in five years.

5. Connect the AI

Install VS Code, then Claude Code, and open your vault folder as the workspace. The AI now reads the same plain files Obsidian does — one folder, two front doors. The full prompt I use is on the one-pager, but the line that matters most is this: before you state a fact about me, find the file it came from, and if it isn't in the vault, say so instead of guessing.

That one rule is the difference between an assistant that knows you and one that invents you.

6. Index it, or none of it works

Two files carry the whole system and they're the ones people forget. An index.md — one line per project with a link — so the AI knows what exists before it starts guessing. And a MEMORY.md with the current state, decisions made and open questions, updated at the end of every session. That's what makes tomorrow start where today ended instead of from zero.

Then run one habit: at the end of each working day, write what you decided and what you got wrong. Two lines is enough. In six months those two lines a day are the only reason you'll be able to look back and see the dots at all.