Loading...
Methodology
I don't just use AI. I design how AI works — so one person can carry the work of a team without lowering the bar.
Every engagement moves through the same three layers. AI accelerates the third — the first two stay human.
Sit where the problem lives. Structure what stakeholders actually mean — not what they first say. Trained in UX research (EQF Level 7), practiced in factories, boardrooms, and cross-functional committees.
Evidence: executive briefings built from primary sources; field visits turned into structured decision material.
Choose what to build, what data model carries it, and what must never be automated. Decisions are recorded, reviewable, and reversible where possible.
Evidence: a 51-table data architecture; a constitutional rulebook for AI, evolved over 30 versions.
A 12-agent AI organization handles research, drafting, verification, and operations under mechanical guardrails. Human time goes into judgment and shipping.
Evidence: 143K lines solo-built in 50 days; production systems with 70+ daily users.
No "should work". Command output, test results, and logs decide what is true.
Handover guides, automated tests, and self-auditing systems — dependence on the builder is a defect.
Adoption beats elegance. Systems live inside the tools people already trust.
Define what AI must never do before expanding what it can do. One rule added means one rule removed.