Most posts about personal knowledge management end with an implicit sell: here is the system that changed my professional life, and you should build one too.
This is not that post.
After three months of building, running, and occasionally breaking Cerebro, here is my honest answer to the question: should you build something like this?
Who This Is Actually For
A second brain is not a productivity tool. It is a relationship management and decision continuity tool. These are different problems.
If your primary value at work comes from execution — from doing clearly defined tasks well and repeatedly — a structured knowledge system has marginal utility. Your context does not compound in the way that makes the overhead worth it.
If your primary value comes from being the connective tissue — from holding context across people, projects, decisions, and time simultaneously, and from making decisions that depend on understanding what is true in multiple places at once — a vault with an AI layer is genuinely useful. The context does compound. The overhead pays off.
The clearest signal: if you regularly walk into a conversation knowing you are missing context you should have, and the cost of that is real, you have a vault problem.
The T-Shaped Requirement
This system requires adjacent skills that are not obvious when you start: markdown, YAML, connected structures rather than linear documents, and enough understanding of how AI context works to make the system intelligent rather than just large.
These are learnable. None of them is deep. But you need to be genuinely willing to acquire them — to spend a weekend writing a skill because the automation matters, to rebuild the vault structure twice before it works, to treat the system as something worth developing.
If that sounds tedious, it probably is not for you. The T-shaped commitment is not optional.
The AI Multiplier
Without the AI layer, the ROI of a structured vault is marginal. Better search, better organisation — incrementally useful compared to whatever you are using now.
With the AI layer, it is a different category of thing. The system becomes capable of reasoning over your history, synthesising across disconnected notes, and surfacing context you had forgotten but had captured. The vault becomes an extension of what you can know, not just a better filing system.
But the AI layer only works if the vault is structured well, maintained consistently, and contains genuine synthesis rather than raw capture. The intelligence the AI can provide is bounded by the quality of what it is reasoning over.
The Minimum Viable Vault
If you want to test whether this is for you before committing to the full system, start here:
- One daily note per day. Three to five lines of what felt important. The discipline of doing it daily matters more than completeness.
- One note per person who matters to your work. What do you know about them? What have they been working on? What is unresolved between you?
- One thinking log. Decisions and the reasoning behind them. Not what you decided — why. One paragraph per decision.
Run those three for a month. If you find yourself consulting them, if they change how you show up in conversations, if the effort feels worth it — add the AI layer. If they feel like overhead for their own sake, stop. The system is not for you.
The Honest No
If you do not write naturally — if turning things into structured text feels like friction rather than clarity — this system will not work. The vault is only as good as the thinking that goes into it.
If you are in a pure execution role where context does not accumulate in the way described above, the overhead is not justified.
If you want something that works without maintenance, this is the wrong thing. The vault requires consistent investment, and that investment pays off only if the underlying habits are there.
Where to Start
Tiago Forte’s Building a Second Brain is the best introduction to the underlying concepts. Obsidian is the tool I use — free, runs locally, and well-suited to the kind of structured vault described here. Nick Zervoudis first showed me what this could look like in practice.
One prompt to try with Claude before committing: give it your last five working days as a set of notes and ask it what patterns it sees. If the output surprises you — if it tells you something about your own week that you had not articulated to yourself — that is the value you are building toward.
If you try it and it does not stick — that is information too.
Credit: Tiago Forte (Building a Second Brain), Nick Zervoudis, Obsidian.
Leave a Reply