Papers and studies

• The same question, three ways →One AI model asked 39 regulatory questions, 36 of them with an answer in the law: from memory, with web search, and with MetLex. What each got right, and how long it took.Study · 8 October 2026
• Answering from the text of the law →Eight places where an answer from memory goes wrong on DORA, FFFS and GDPR, and the method that catches them.Being revised; ask for a copy
• What AI-assisted compliance work costs →What the cost of AI buys, and what stays with people.Being revised; ask for a copy

Talks

• AI that cites the law →What changes when an AI answers from the text of the regulation.MetLex, 2026
• Building with an AI coding agent →What the agent writes, what a person must check, and where it goes wrong.TermoSense, 2026