Intermediate
Build Your AI Second Brain
Learn Andrej Karpathy's LLM Wiki pattern: a self-improving knowledge base where Claude builds and maintains a structured wiki from your sources so knowledge compounds over time instead of evaporating between sessions.
01
The Problem with How We Store Knowledge
Why notes apps fail, why RAG falls short, and the core insight from Andrej Karpathy that makes an LLM-maintained wiki fundamentally different from every productivity system you've tried before.
›
02
How the Second Brain Thinks
The three-layer architecture behind the LLM Wiki pattern — raw sources, wiki, and schema — and how the self-improving loop turns every source you add and every question you ask into compounding knowledge.
›
03
The Architecture: Folders, Files, and Rules
The exact folder structure, the two hard rules that make the system work, and what goes inside CLAUDE.md — the operating manual that turns Claude into a disciplined wiki maintainer rather than a generic chatbot.
›
04
Set It Up Today (Claude Code Walkthrough)
A complete step-by-step setup guide: creating the folder structure, writing your CLAUDE.md operating manual, scheduling the weekly Dream Sequence, and running your first source ingest from a URL.
›
05
The Five Operations in Depth
A detailed walkthrough of all five operations: Ingest, Query and the save-that workflow, Dream Sequence, Index and Log, and Session Capture. These five operations are everything you need to keep a knowledge base compounding.
›
06
Making the Most of Your Second Brain
Advanced tooling and integrations: Obsidian as your wiki browser, the Web Clipper for one-click source capture, qmd for search at scale, and git as automatic version history for everything the LLM builds.
›
07
Keeping It Alive: Patterns That Stick
The common failure modes that kill personal wikis, the minimal viable workflow that avoids all of them, and what a second brain that has been running for six months actually looks like.
›