Building Your Second Brain
I want to put something in your hands this week that I genuinely believe is one of the highest-leverage things you can build right now.
A second brain — an AI-powered knowledge base that reads everything you feed it once, then builds and maintains a living, interconnected library of your knowledge for you. You stop re-explaining yourself to AI. It already knows your work, your projects, your thinking — and it gets smarter every time you add to it.
This is Andrej Karpathy's "LLM Wiki" idea, and here's why it matters for you specifically.
🌍 Why this matters as you transition into the workforce
The job market isn't rewarding people who can use AI. Everybody can open a chatbot. It's rewarding people who can architect with it — who build systems rather than ask one-off questions.
That's the whole difference between "a prompt engineer who asks ChatGPT for code" and "a system designer who uses AI as a force multiplier." One is replaceable. The other gets hired.
When you build your own second brain, three things happen:
🧩 You demonstrate AI fluency — the exact skill every forward-looking employer is screening for. You don't just talk about it; you show it.
📈 Your knowledge compounds. Every note, transcript, and lesson you add makes the whole thing more valuable. That's an asset that grows while you sleep.
🎯 You stand out. Most people consume AI. You'll be one of the few who design with it. That's the portfolio piece and the interview story.
This isn't theory. I built mine over the last few days, and it now holds 600+ pages of my entire body of work — and I can talk to it from Slack like a coworker.
🛠️ What you're building (and the part I want you to hear)
Three pieces:
- Obsidian — a free app that holds your knowledge as simple text files. This is your "vault."
- Claude — the AI that reads your stuff and builds & maintains the wiki for you. You barely touch it; the AI does the bookkeeping humans always abandon.
- An agent (connected through Slack or Telegram) — so you can ask your second brain anything, from anywhere, anytime. Mine's named Blake. He lives in my team channel and answers from my brain 24/7.
👉🏾 Here's the part I really want to emphasize: you can build this entirely inside the Claude Desktop app.
You connect a folder, talk to it in plain English, and it does the work. The setup videos below show Claude Code in a terminal — but everything they do, you can do by just describing it in the Claude desktop app. Don't let the terminal scare you off. If you can have a conversation, you can build this.
▶️ Your resources (read first, then watch)
1. Read this — the original idea (start here so you understand the why): Andrej Karpathy's "idea file" for the LLM Wiki
2. Watch this — full beginner setup, no coding needed: "Karpathy's LLM Wiki — Full Beginner Setup Guide" (Teacher's Tech). Walks you through the Obsidian vault, the schema, ingesting your first doc, and keeping it healthy.
3. Watch this — the 5-minute version + real examples: "Andrej Karpathy Just 10x'd Everyone's Claude Code" (Nate Herk). Same idea, fast, and he shows two of his own second brains, so you see what's possible.
🚀 Your move this week
Start small. Make one Obsidian vault. Drop in your notes from this program, a few articles you've read, even your old project files. Open the Claude desktop app, point it at that folder, and ask it to build your wiki. Watch it connect things you'd forgotten you knew.
Then come back here and tell us:
- What did you put in your brain first?
- What's one connection that surfaced that surprised you?
- Where are you stuck? (Drop it below — we build together.) 👇🏾
Remember the throughline of everything we do here: AI is empowerment, not replacement. A second brain isn't about outsourcing your thinking — it's about freeing you to think bigger, while the AI handles the remembering. That's the posture that's going to carry you into this next era of work.
Build it. Show me what you create.
⬇️ Check out my exact prompt that I entered to start this project in the comments ⬇️