Your Second Brain Is Just Files in a Folder

Where the phrase comes from

The term got popular through a man named Tiago Forte and his method, Building a Second Brain. The idea is older than him, people have kept external notes for as long as there have been notes, but he gave it a shape that others could follow. His version has two parts worth knowing.

The first is a workflow he calls CODE: Capture, Organize, Distill, Express. You save the things worth keeping, arrange them, and boil them down to something you can actually use later. Then the result becomes output, which is the whole point of keeping any of it. The second part is how you file things, which he calls PARA: Projects, Areas, Resources, Archives. The trick there is that you organize by what something is for, not by what topic it belongs to. A note about color theory does not go in a folder labeled “design.” It goes with the project you are using it on.

For years that was the whole game. You kept your second brain in an app like Obsidian or Notion, you linked your notes together, and when you needed something you went and found it yourself. The brain was a library, and you were still the only one reading it.

What changed

Your second brain is just files and folders on your computer, organized and named in a way that makes sense, mostly plain text. Notes you wrote, documents you saved, papers you scanned in. That is it. There is no special software you have to buy and no degree you need. If you have a folder with documents in it that are arranged well, you already have the skeleton of one.

What is new is who else can read it. An AI can now sit on top of that folder and use everything inside it. So instead of you digging through your own notes, you ask, and the AI answers using your material, not the open internet, just the things you decided were worth saving.

I want to correct one thing while I am here, because it trips people up and I said it wrong myself for a while. People say they “trained the AI” on their notes. Almost nobody is actually doing that. Training a model means changing how the model itself works, which is slow and expensive and not what happens when you point Claude or a custom GPT at your files. What is really happening is simpler. The AI reads your files fresh each time, the way you would open a book to look something up. Nothing gets baked in. You change a file, and the very next answer reflects the change. The cleanest way I have heard it put is that training is a closed-book exam, and this is an open-book one. The AI is not memorizing your knowledge. It is consulting it.

Tools that work this way right now: Claude Projects, where you load files into a project and it draws on them across every chat. Custom GPTs, where you attach knowledge files. Google’s NotebookLM, which only answers from the sources you give it and nothing else. And the one I live in, Claude Code, which reads the actual files in my projects plus a set of memory files I keep, so it walks into every session already knowing my rules, my brands, and my history.

Why this is worth your time

Two reasons, and they are the real payoff.

The first is your skill set. Every time you save something you understood, a fix that worked, a process you figured out, a decision and the reason you made it, you are building a record of how you think. The AI can read all of it at once, which you cannot. You hold maybe a few things in your head at a time. Your folder holds everything. So when you are stuck, the answer is often already in there, written by a past version of you who solved this once before. The second brain turns your own experience into something you can ask questions of.

The second is new ideas, and this is the part I find genuinely moving. When an AI can see across all of your material, it can notice things you would never connect on your own, because you were never looking at all of it on the same day. A note from January and a problem from June. A pattern across three clients you handled months apart. You ask it to find the thread, and sometimes it hands you a concept you did not have before, built entirely out of pieces you already owned. It is not inventing from nowhere. It is showing you what was always sitting in your own files.

How I actually use it

Most of what I do with AI is not asking a question and reading the answer back. It is agentic, which means I point the AI at my files and tell it to go do work, and then it does. Not “explain this to me,” but “take everything I have on this and build the next thing.”

And it builds. This is the part people still underestimate. From the same folder of notes, I can have it write a working app, not a description of one, the actual thing, running. I can have it stand up a workflow that runs on its own, wire together an automation that handles something I would otherwise do by hand, or make a small tool that did not exist that morning and exists by the afternoon. The knowledge base is the raw material. The AI is the set of hands that turns it into something that actually runs.

That is where the second brain stops being storage and becomes a collaborator. I can tell it to read across months of notes on one theme and pull them into a single connected idea. I can tell it to take a concept I have been circling and cement it, find every place I have touched the idea before, and turn the scatter into something solid. Or I hand it the raw material and ask it to make something new from it, a draft, a plan, a workflow, an app, built out of my own knowledge instead of the generic middle of the internet.

The difference is real. A second brain you only read is a better filing cabinet. A second brain you can direct is closer to working next to another version of yourself, one that has read everything you have ever saved, never gets tired, and can build.

Last week I was scanning papers, sorting files, putting things where they belong. I am still doing it. That is not the boring part of the work, it is the work. Scanning and filing what you know is how a second brain gets built in the first place, and the more carefully it goes in now, the more the AI can build with it later.

That is the whole thing. A second brain is a place to set down what you know, a little at a time, so an AI can pick any of it up and make something with it whenever you point it there.

So start one. One folder, today. Put the first thing in it that you do not want to lose, then the next, then the one after that. It will feel like nothing for a while. Then one day you point an AI at it and watch it build something out of years of you, and you finally see what you were doing all along.

You just need a folder and the decision to keep what matters. And if the privacy of it worries you, remember that local AI is a thing now. There are models that run entirely on your own machine, so your second brain never has to leave your computer. Open it tonight. Start your brain.

Forward → Upward ↑ Onward ↗︎
Mstimaj

Sources and Further Reading

• Forte Labs: Building a Second Brain, An Overview (the CODE workflow and the origin of the method).
• The PARA Method (Projects, Areas, Resources, Archives).
• NVIDIA: What Is Retrieval-Augmented Generation? (plain-language explainer for how an AI reads your documents).
• IBM: RAG vs. Fine-Tuning (the difference between reading your files and actually training a model).
• Anthropic: What Are Projects? (loading your own files into Claude as a knowledge base).
• Google NotebookLM (an AI that answers only from the sources you give it).