# Building a Second Brain with AI: What I Learned After Three Months
My system was not designed for the role I was actually doing.
I am the Chief Product and Technology Officer at Wall Street English, a language education company operating across 70+ countries. My working day spans Barcelona and Chennai — a 4.5-hour time difference that makes almost everything asynchronous. I manage teams across two continents and interact regularly with markets across Europe and Asia.
The job is not primarily execution. It is holding context across people, decisions, projects, and time zones simultaneously — and connecting them in ways that the people inside each of those contexts cannot see from where they sit. That is what strategy means at the level of daily work.
For years, I managed this the way most senior leaders do. Notes in Notion that became a graveyard. Bookmarks I never revisited. Meeting summaries that captured what was said without preserving what I understood. It worked, loosely, until a 1:1 where I arrived without context on something the other person had been thinking about for weeks and expected me to have held.
That was the moment I decided to build something different.
What I built
What I built is a connected knowledge system: a structured vault of notes and relationships in Obsidian, with an AI layer — Claude — that can query and reason over everything in it.
I called it Cerebro. The name is accurate. It functions as a second brain — not storage, but memory. A knowledge base without reasoning is an archive. Reasoning without a knowledge base starts from scratch every time. Together they do something neither does alone.
Building it demanded skills I did not expect — not technical ones, those were straightforward enough, but a clear mental model of what I was actually trying to solve, the discipline to build incrementally rather than design the perfect system in advance, and a willingness to change how I work, not just bolt a tool onto how I already work. What I actually built, and the skills it demanded →
Why senior leaders need this specifically
The context-switching cost is not evenly distributed. It is highest for the people who are structurally the connective tissue of an organisation.
A senior leader’s value is in holding threads simultaneously that no one else can see from a single vantage point. When you lose context between a Chennai 1:1 and a Barcelona leadership meeting, or between a decision you reviewed last month and one that turns on it today, you are not just slower. You are structurally worse at the thing your role requires.
Most knowledge management tools are built for capture. They solve for getting information in. They do not solve for surfacing the right information at the moment you need it. That gap is what a second brain, properly built, closes. The context-switching tax no one talks about →
What actually changed
Three months in, the changes that mattered were not the ones I expected.
The time saving is real but not the most significant outcome. What changed more substantially was the quality of decisions that turn on context held across time or across multiple people’s situations simultaneously. I show up to conversations differently — not because I have more information, but because I have the right information retrieved rather than reconstructed.
The other shift I did not predict: I use it as much for preparation as for retrospective capture. Before a significant conversation, I surface what is relevant. That changed how I enter those conversations, which changed what they produced. Three months in — what actually changed →
What I got wrong
The biggest mistake: I built too much structure at the start.
A connected knowledge system needs enough structure to be navigable, but the structure has to emerge from use, not be designed in advance. The parts of the system that work are the parts that started simple and grew. The parts I designed fully before using are the parts I abandoned. Gall’s Law applies here directly — a complex system that works is invariably found to have evolved from a simple system that worked — and I ignored it.
I also still get things wrong. The system has not made me perfect at knowledge retention. It has made me specifically better at a specific set of things, and revealed new failure modes I did not have before. What I got wrong, and still get wrong →
Should you build one?
Not necessarily.
Most posts about personal knowledge management end with an implicit sell. This one does not. Whether a second brain is worth building depends on specific things about your role, your working patterns, and what problems you are actually trying to solve.
The question is not whether a second brain is a good idea in general. It is whether your role has the characteristics that make the investment worth the cost — and whether you are willing to change how you work, not just add a system on top of how you currently work. My honest criteria for whether you should build one →
The Second Brain Journal — five posts on building a personal knowledge system with AI:
– The Context-Switching Tax No One Talks About
– What I Actually Built (And The Skills It Demanded)
– Three Months In — What Actually Changed
[…] Series guide: Building a Second Brain with AI […]