Physician and enterprise product leader. Author of the Agentic AI for Product Leaders series. Writes on AI, healthcare, and what changes when software starts acting. 15+ years at SAP, Walmart, and regulated health tech.
I tested eleven technologies and eight LLMs on one triage problem over a weekend. A logistic regression won. The bigger finding: a test that used to take a quarter now takes a weekend, and that changes what an architect is for, especially where the worst case is a missed diagnosis.
A relative lost his job to an agent, so I built him one. Choosing which model should interview him cost a day, $22, and one reversal I did not enjoy. The model was never the problem. Six other things were, and finding them is the eval nobody runs.
The brain processes the world at walking speed. Nobody processes 120 mph, and nobody processes a coding agent. Sport riding taught me the only technique that works: decide everything before the ride, then look at the exit, never at the pothole. The agent, like the bike, goes where you look.
A coding agent asked a product question at 10:40 on a Tuesday. The PM was in a review, the designer was on another project, the developer was at lunch. The agent took the default, the tests passed, and the product decided for itself. Who answers the agent? Everyone, by topic, with a default.
Watching is not a job. It is the transitional state between two jobs, and it is the most dangerous place in the building, because it looks like supervision and it is not. I hired a team on Saturday. By Sunday it did not need me, and I was there anyway, pressing the green button.
Two frontier AI models, an external reviewer, thirty-nine signed rulings, and not one of them ever said stop. I did not fire my agent team for being wrong. Every answer was right. I fired them because no answer ever included the sentence that mattered.
I am working with the best developer of my career. Zero latency, infinite patience, always says yes. It is also the most dangerous working relationship I have ever had. On what happens to product judgment when implementation becomes free, and where the discipline has to live now.
Ten runs of one boring question about fruit teach more about AI than any slide deck. Probabilistic is not a definition; it is something you watch happen to your own keyboard.
Physicians are building their own AI tools, and patient data cannot leave the building. So which local model deserves to be in the room with a chart? Not the biggest one. I measured.
A PM who has never watched an agent work cannot write its spec, its evals, its boundaries, or its intent. The pod model assumes PM, UX, and engineering share an understanding of the product's behavior. For agentic products, only one of the three has it.
Data products were built on a quiet assumption: the designer could anticipate the questions. Agents break that assumption without breaking the discipline. What survives is not the shape we prepared but the honesty of the contract underneath it. Five questions for every data product owner.
Healthcare asks who is accountable for a machine's mistake before it asks what the machine can do, because it has no choice. On 18 August a federal regulator applied that question to generative AI. The framework it produced is not new. It is clinical governance, written down for everyone else.