What an AI Second Brain Actually Is
A lot of people are talking about "agentic OS" and "second brains" without ever defining them. In this walkthrough I pin down exactly what an AI second brain is, in two concrete levels: level one, connecting your everyday tools to AI so it answers from your real context, and level two, the LLM wiki, an index that lives on your own computer and links everything you do. By the end you will know precisely where to start and what each level buys you.
What you need
- An AI application like Claude, ChatGPT, or Gemini.
- The tools where your life actually happens: your calls, emails, CRM, and documents. Level one is just about connecting them.
1. The problem: AI does not know your world (0:26)
Whenever you use an application like Claude, ChatGPT, or Gemini, you are talking to a huge brain that holds a lot of the internet's information. The problem: it does not know anything about your business or your personal life. Our real knowledge, the things we do daily, lives in external tools: calls, emails, CRMs, documents. None of that is inside the AI brain by default. That gap is the entire reason second brains exist.
2. Level 1: connect your tools to AI (0:51)
The first step is connecting the dots. You take all the tools you use every single day and connect them to the AI. Now the AI has access to knowledge about you and about what you are doing, so when you ask a question, you get a precise, robust answer that comes from exactly that knowledge instead of generic internet text.
That is level one of an AI brain: just connecting it to the right tools so it gives you good answers on anything you ask. Do not underestimate it. It will get you maybe fifty to sixty percent of the way there, and you will be shocked at what you can do with this level alone.
3. Level 2: the LLM wiki, or Obsidian brain (1:31)
Level two is what you mostly see on YouTube: the Obsidian brain, or the LLM wiki. Why does it exist? Because level one has a weakness: we do things and we do not update our systems as much as we should. You send an email to a prospect and never log it in the CRM. You finish a call and never write down what was said. Information ends up laying around everywhere all at once, and that is a huge issue, because connected tools can only answer from what is actually in them.
The fix: instead of just connecting AI to the tools, you ask the AI to read everything inside them. It reads through your emails, your CRM, the documents of your business, or simply the things you enjoy, and it establishes connections in what we call an index.md. Do not be scared by the .md part; it simply means a text file.
4. Building your index.md (2:52)
Here is how the index works in practice. Imagine you have an email with person A, and you also have a call with person A. The AI reads both documents, the call transcript and the email you sent, notices the connection between them, and records it in the index. The index becomes the document that lists out all of your clients, all of your processes, your offers, your decisions.
And every time you add a new email, the AI reads through it again and folds it into the index. You are updating that brain each and every time, so nothing goes out of sight of the AI.
5. What goes in your second brain (3:29)
The beauty of a second brain is that you are the one who decides what to build. You choose what goes into it, so it is shaped around how you work. You do not have to change the way you work to build this. That is the benefit of level two: pretty much everything you do in daily life becomes accessible from a single application, whether that is Claude or Codex, it does not matter.
There is a strategic bonus too. Because the level two brain lives inside your own computer as plain text files, it is portable. If there is ever a problem with Claude, or you want to migrate to another AI application, you simply take your index and go. No lock-in.
6. Setting up level 1: it really is simple (4:40)
Setting up level one is extremely simple: go into your Claude account and connect it to the tools you use on a daily basis. That alone is largely enough to start, and it will show you what having an agentic OS feels like, at least at level one. Then, when you get more advanced, you move on to level two and bring the brain onto your own machine.
7. Bonus: a second brain for your books (5:42)
This is not only for business. If you have a bunch of books sitting on your computer, ask the AI to build a second brain out of them, to see how they match each other and how ideas link together. It is fantastic for brainstorming and research, and a low-stakes way to practice the whole approach on a subject you love.
Pitfalls and tips
- Do not stop at level one and assume you are done. It gets you fifty to sixty percent of the way; the unlogged emails and calls are exactly what it misses.
- Do not reorganize your life to fit the system. The second brain is shaped around how you already work, not the other way around.
- Keep portability in mind. An index living on your computer means you can switch AI applications whenever you want.
Where to go next
A second brain pairs naturally with repeatable workflows: build one in How to Build Your First Claude Skill, then let it run without you in Automate Your Work with Claude (Scheduled Tasks).
