I want to understand how things work.

Show me anything and I want to follow it down. An AI model runs on GPUs, the GPUs come out of a chip plant, the plant runs on what a mine digs up. I want the whole chain, and I want to know who does the work at each step.

That has dragged me through law, physics, economics and history without letting me settle in any of them. I trained as a lawyer, then left for Germany to study physics because I wanted to reach general relativity. The math was harder than I was. I am still at it.

Until I was about twenty-five I thought history had nothing to teach anyone. Then a few strategy games showed me economies as systems with moving parts, and I never got back out. It turned into a novel: 2,400 people from present-day Shanghai land in southern Africa in 1628 and try to build a country. I wanted to know what that would really take, down to who makes the toilet paper. The first twelve days ran to close to a million words, and the world underneath them is now a model that refuses to compute when the world I describe is impossible.

I learn a subject by building something about it. Economics became an interactive textbook, walkthroughs of the questions economists still argue about, timelines and atlases. That is Uniflection.

For a living I build and run businesses. I spent seven years running a trading and sand-dredging company on the ground in Cameroon, from nothing to about fifty people. Then hardware brands in Shenzhen, other people's and a consultancy of my own. Since 2026, a consumer AI company with an old friend. What I took from all of it is that knowing what to do is the small part. Getting it done, with the people and the machines you actually have, is where the difficulty lives.

For about ten years I tried to learn programming and it never stuck. In 2025 AI started writing the code, and it finally did. Agents now do the operating work under everything above, this page included.