Source: This is a summary of a YouTube video featuring Greg Eisenberg and his guest Vin (Internet Vin). Watch the original video
Greg and Vin walk through how pairing Obsidian (a plain-text note-taking app) with Claude Code (an AI agent) creates something more powerful than either tool alone — a personal thinking system that notices patterns in your life, helps you reflect, and turns your notes into real actions. The conversation moves from setup basics all the way to live demos of AI reading someone's personal vault and generating surprisingly personal, useful ideas.
Claude Code is an AI agent you run from your computer's terminal that can read, write, and manage files using plain English. Instead of re-explaining a project every time you open a new chat, you save that context as a file and simply point Claude to it. The core insight: the quality of what AI can do for you is entirely determined by the quality of context you give it.
Obsidian is a free app that sits on top of a folder of plain Markdown (.md) text files — your "vault." What makes it different from a regular folder is that you can link files to each other, the way memories connect in your brain. A note about a meeting can link to a note about a person, which links to a project — and over time, these connections build a map of how you actually think.
The Obsidian CLI (a plugin) lets Claude Code not just read your notes, but also understand the relationships between them. This is where things get interesting: Claude can spot a pattern you've been circling around for a year without realizing it — and naming that pattern can be a genuine breakthrough.
Vin demonstrates several custom commands he created (which you can ask Claude Code to build for you):
/context— loads a full picture of your life, work, and current state before you start any session/today— pulls your calendar, messages, and recent notes into a prioritized plan for the day/close day— end-of-day review that surfaces action items and checks your outstanding hypotheses/ghost— answers a question the way you would, based on your vault's writing style/challenge— pressure-tests your current beliefs using your own past writing to find contradictions/emerge— surfaces ideas your notes imply but you've never quite stated out loud/drift— compares what you say you want to focus on against what you've actually been doing over 30–60 days/ideas— deep vault scan to generate ideas across all areas of your life (tools to build, films to watch, people to meet)/trace— tracks how a specific idea has evolved in your thinking over months or years/connect— takes two unrelated domains (e.g. filmmaking and worldbuilding) and finds the hidden bridges between them
Vin's favorite use isn't building software — it's thinking. He uses Claude to help him develop ideas, challenge his assumptions, and see himself more clearly. He keeps a strict rule: he writes everything himself; the AI reads and responds, but doesn't write into his vault. This keeps the vault a true reflection of his thinking, not the AI's.
Writing isn't just journaling — it's how you generate ideas and delegate to AI more effectively. The more you write, the richer the context you can hand off. Vin notes that as adults, we tend to write and reflect less the busier we get, right when we could benefit most from it.
Greg draws the analogy to therapy: a good therapist doesn't tell you what to think — they guide you to surface things yourself. This system does something similar. It helps you see patterns and connections that are genuinely hard to spot when you're living inside your own life.
The live /ideas demo produces a surprisingly specific personal report — not generic advice, but suggestions like: build this specific tool, have a conversation with this person, write this essay, investigate this subject. It even suggests people Vin should meet. The output is rooted in months of his actual writing.
- Your notes are your leverage. The more clearly you write down what you're working on, thinking about, and caring about, the more an AI can actually help you — not with generic advice, but with guidance that fits your life.
- Stop re-explaining yourself. Write your goals, projects, and preferences into files once, then reference them. This alone saves enormous time and makes AI feel far more useful.
- Reflection isn't passive. When AI can read your past notes and surface a pattern you've been circling for months, it can accelerate your thinking in ways that feel almost unfair.
- The vault should be yours. Vin deliberately keeps AI out of his notes — it reads but doesn't write. This preserves the vault as a true record of his thinking, not a mix of his ideas and the AI's.
- Writing is now how you delegate. A clear written description of a project or preference is something you can hand to an AI agent, now or later. Good writing habits have a new, practical payoff.
- You only need to explain things once. Unlike a human colleague who might leave or forget context, a file stays. You build up a base of context over time that keeps getting more useful.
If you want to try some version of this without the full setup:
- Start a simple notes habit — even one file where you write down what you're working on and what's on your mind. Plain text or Markdown is ideal. Don't worry about making it pretty.
- Download Obsidian (free, open source) and create a vault — just a folder of text files to start. No need to configure anything elaborate at first.
- Try pasting a chunk of your notes into Claude (web version is fine to start) and ask: "What patterns do you notice in how I'm thinking about X?" You don't need Claude Code to begin getting value from this.
- Write a single "about me" context file — your current projects, what you care about, how you like to work. Paste it at the start of any AI conversation. Notice how much better the responses get.
- Try the
/driftidea manually — look back at what you said you wanted to focus on a month ago and compare it honestly to what you actually spent time on. Often illuminating, no software required.