Physician · Product Leader · Author
What it takes for AI to successfully move from pilot to production, in rooms where being wrong has consequences.
The thesis
“The team ships the agent. The product fails because no one ships the system that supervises it.”
From Why Agentic AI Products Fail
About
Medicine trains you to act under uncertainty, insist on evidence, and own the outcome when a decision is wrong. That instinct is what enterprise AI product managers need as software starts to act on its own.
With 15+ years building enterprise cloud and data platforms, I pair the product track record of a senior technology leader with the judgment of a trained physician. The clinical background isn't a domain specialty; it's a lens: rigorous about evidence, deliberate about supervision, and clear-eyed about what it means when a system acts before anyone reviews the decision. I bring that lens to agentic AI wherever being wrong has consequences: operations, finance, enterprise data, and healthcare alike, shipped at Walmart scale, SAP enterprise, and earlier at Microsoft.
Senior Director at Walmart Global Tech ($400M+ in operational savings; a 10-person global PM team). Most recently leading product areas for SAP Business Data Cloud, SAP's data foundation for enterprise AI, and previously shipping regulated clinical health tech at Hello Heart.
The hardest part of technology transformation isn't the architecture; it's the people. I focus on the human and cultural side of change: how teams adopt new tools, where resistance comes from, and how to design rollouts that respect workflow reality and build genuine trust. Technology that isn't adopted isn't a product. It's a cost.
Recent · Harvard Medical School / Emeritus
Work
A four-book series on building agentic AI products, and keeping the human judgment they route through. A standalone book on what medicine is risking by letting AI into the room. And 100+ articles, with new writing weekly.
Standalone · On medicine
The Second One in the Room
Medicine let AI into the room and looked away, and almost no one noticed that the wall keeping watch went with it. A field report from that dissolution.
Available on Amazon ↗
Articles · Weekly
Data, Decisions, and Clinics
100+ articles on healthcare AI, enterprise data architecture, regulation, and what it takes for AI to move from pilot to production in a regulated environment.
data-decisions-and-clinics.com ↗
Companion · Skill package for Claude Cowork
Agentic AI PM Skill Package
Seven skills covering the agentic PM lifecycle, distilled from the series. Invoke them at the moment of need.
On the shelf
The signature framework
Six phases that name what a product manager actually does when shipping an agentic AI product. The backbone of the book, the article series, and the SAP Community posts. Most teams enter Phase 1 without Phase 0. That's where the failures begin.
Deliverables
Questions answered
Phases 4 and 5 form a continuous feedback loop. Observe is the measurement layer; Operate is the response layer.
Frameworks
50+ named mental models developed across the writing, each with an insight, a key phrase, and a source. The vocabulary that makes the lifecycle operational. Nine shown.
{{ f.num }}
{{ f.name }}
{{ f.phrase }}
Nine of 50+ shown. The rest live in the book and articles. Browse the full library →
Speaking & impact
Most people write about AI. These come from someone who trained in the clinical rooms AI is now entering, and then spent fifteen years building the enterprise platforms displacing them.
{{ b.label }}
{{ b.headline }}
{{ b.text }}
Available for selected speaking engagements, podcasts, and advisory discussions on agentic AI product design, AI governance in regulated environments, and what medicine is risking by not watching closely enough.
Get in touch
For speaking, advisory, workshops, podcast invitations, or to share a hard problem in healthcare AI or enterprise product. I respond.