Every so often an idea with a grand name turns out to be something plain. A second brain is one of those. Strip away the branding and it is one place where what you know is written down, kept current, and organised well enough that someone who is not you could pick it up and use it.
For most of my working life that place did not exist. Context lived in my head, in old emails, in chat threads, in project folders named final_v3. Every tool held a fragment. Nothing held the whole.
Then AI became good enough to do real work, and the gap started costing me money.
The problem was never note-taking
I had taken notes for years. Most people have. Notes were never the problem. The problem was that nothing I wrote was written for anyone but me, in the mood I was in that day, in whichever app was closest. That is storage, not memory. Storage holds things. Memory hands you the right thing at the moment you need it.
AI exposed the difference quickly. A model with no context is a brilliant stranger. Every session started from zero: who I am, what my businesses do, how we write, what we decided last month and why. I was spending the start of every working session re-teaching a tool my own life, then watching that teaching evaporate when the chat ended.
The fix was not a better model. The fix was a written home for the context the models kept asking for.
Plain text, boring on purpose
My second brain is plain-text files in ordinary folders. No clever app at the centre, no database, nothing that needs a subscription to open.
That choice looks unambitious, and it is the most important one I made. Formats with the least ambition survive the longest. A text file written thirty years ago still opens today; the shiniest tool of that era does not. And everything can read a text file, which matters more now than it ever has, because the tools change every few months and the brain should not.
When people ask which software to use, they are asking the wrong first question. The software is the least durable part of the system. The writing is the system.
Write for a reader with no memory
The habit that changed the most for me: every note is written for a reader who knows nothing about today. Usually that reader is me in six months. Increasingly it is an AI assistant five minutes from now. The discipline is the same either way. Say what is true. Say why the decision was made, not just what it was. Name things properly instead of pointing at them.
The test of a note is whether someone could act on it without asking me a question. Most notes fail that test the first time you apply it. Mine did.
What AI changes
Given the brain, an assistant stops being a brilliant stranger and starts being useful in a way that compounds. Drafts come back in my voice, because the voice is written down. Briefings arrive already knowing what happened last week, because last week was recorded. Work starts from what is actually true rather than from a confident guess.
The lesson from living with this daily: the quality of AI output tracks the quality of the written context far more than the choice of model. Everyone is arguing about which model is smartest. Very few people are writing down what any model would need to know to be useful to them specifically. That gap is the opportunity.
The rules matter more than the tools
A system like this earns trust the way a person does, slowly and by behaving well under pressure. So it runs on rules that do not bend.
Everything an assistant produces is a draft, and a person approves anything that leaves the system, every time. Missing information stays blank instead of becoming a confident guess. Personal, financial and client information is kept apart, and any tool sees only what it needs for the task in front of it. The most sensitive work runs on my own hardware, where nothing is retained by someone else's product.
None of this is exotic. It is the same judgement you would apply to a new employee with access to everything. You would not let them send email on your behalf in week one either.
The blueprint stays private
I write about the principles and keep the internal layout to myself, deliberately. Publishing the floor plan of a system that holds your life and your businesses is not transparency, it is negligence. The principles travel. The blueprint should not.
If your second brain is worth building, it is worth protecting like the things you put in it.
Start smaller than feels serious
If I were starting again I would not start with software. I would write one honest page about the work in front of me: what it is, what done looks like, what I have decided so far and why. Then I would keep that page true for a month. That is the whole habit. The folders, the structure, the AI, all of it grows from one page kept true.
The second brain is not the impressive part of anything I have built. It is the quiet part that makes the impressive parts work.
A companion note covers the other half of this idea: what changes when a business, rather than a person, writes down what it knows.