PRODUCT THINKING
How I Think About Enterprise AI Products
Principles I use to build enterprise AI products that are useful, governable, and adopted in real workflows.
Discovery before roadmap, always.
I start with structured customer interviews, churn cohort analysis, usage patterns, and competitive mapping before writing requirements. Most product failures are misdiagnosis problems, not execution problems.
Workflow first, model second.
Before designing agents or AI features, I map the decisions, handoffs, exceptions, controls, and business outcomes that shape the workflow. Enterprise AI only works when it understands the operating reality.
Autonomy needs trust infrastructure.
I design human-in-the-loop controls, observability, explainability, cost governance, and auditability into AI products from day one. Enterprise buyers won’t deploy what they can’t monitor, explain, or govern.