
November 23–27, 2026
Product Leader and Systems Analyst
Independent
AI agents can complete tasks yet still leave product teams unable to trust the result. This talk shows a practical operating system for product work with agents: a product question becomes a bounded task, evidence is captured, agent outputs are evaluated against explicit acceptance criteria, corrections are recorded, and a named human accepts consequential decisions. I will walk through non-sensitive artifacts from my own product projects, including a failure mode and the resulting change to the workflow. Attendees will leave with a reusable decision boundary, evaluation checklist, and way to move from promising demos to accountable day-to-day work.
See every session across the AI Agents Conference November 2026 line-up.
View Agenda