Building My Second Brain
Since reading Tiago Forte’s book Building a Second Brain, I’ve been hooked on the dream of one repository of knowledge personal to me. I’ve been chasing how to organize all of the content I’ve been reading, writing, and collecting over the years. I read this book around the time ChatGPT came onto the scene, and I was fascinated to see how I could feed content into this new tool to help me with my work.
“This digital commonplace book is what I call a Second Brain. Think of it as the combination of a study notebook, a personal journal, and a sketchbook for new ideas.”
— Tiago Forte, Building a Second Brain
Commonplace Books and the 4 Purposes
In his book, Forte outlines the rich history of commonplace books throughout history, highlighting many notable figures who kept their own. He outlines four purposes of a second brain that I think are critical to anyone who makes things:
“There are four essential capabilities that we can rely on a Second Brain to perform for us: Making our ideas concrete. Revealing new associations between ideas. Incubating our ideas over time. Sharpening our unique perspectives.”
— Tiago Forte, Building a Second Brain
The LLM Wiki Insight
Recently, I found an echo of Tiago’s ideas in Andrej Karpathy’s llm-wiki · GitHub. He suggests using simple markdown files with tagging and linking to create a sort of knowledge graph specific to you—feeding this wiki into LLMs so they could learn with you as you solve problems together instead of always starting from scratch.
This simple idea clarified the multi-year journey I’ve been on and gave me better direction.
My Current Second Brain Stack
My current second brain stack consists of:
- Obsidian — To store notes in simple, text-based markdown with links and tagging.
- Readwise — To capture highlights (synced directly to the Obsidian vault).
- AI Agent Assistants — Tools like Antigravity, ChatGPT Codex, Claude Code, Hermes, or OpenClaw.
Surprisingly, all the agent harnesses seem to be able to easily treat an Obsidian vault synced with git as a first-class knowledge repository. They can handle the linking seamlessly.
Just as Karpathy and Forte suggested, the LLM and digital tools start organizing the data. My data remains portable, organized, and connected by topic through tags and backlinks.
What’s Next
While I’m early in the journey, I think it is worth capturing how this new knowledge stack works and details about how each piece of the stack operates together. I’ll dive deeper into the details in future blog posts.