Alexander Sucala
Alexander builds AI systems that either prove their answers are correct or refuse to answer. 30+ patent filings. 14+ published papers. 4 production systems running right now.
What Alexander does
Most AI systems have a dangerous habit: when they don't have enough information to answer correctly, they guess and present the guess as if it were fact. The output looks confident. It might be completely wrong. And you can't tell the difference until real damage is done.
Alexander builds systems that make this impossible. If the AI can't prove its answer from real data, the answer doesn't ship. If the math says one thing and the AI says another, the math wins. Every finding is labeled with where it came from so you can tell the difference between something backed by your actual documentation and something the AI thinks is a good idea.
These aren't research prototypes. There are four systems running in production right now, each one implementing the same patented architectural patterns across different domains: automated code review, financial analysis, autonomous trading, and daily intelligence briefings. The Systems page explains how they work. The Patents and Papers pages document the ideas behind them.
Outside of work
Motorcycles, hiking, travel, reading, and building things that don't break.