Practitioner POV on shipping clinical products, AI in product orgs, and the moments where founder-led product breaks. Cross-published from LinkedIn — the original discussion lives there.
Patient and provider lenses tell you if a feature should exist. The ops lens tells you if it survives contact with reality — whether the admin can see what the logic does, change it, or turn it off without an engineer in the room.
Read →Origin's clinicians had outcomes data that wasn't accessible during patient visits. Buying access would have run $250-400 per provider per year across 110+ providers. I built a Chrome extension on top of their EMR instead — six weeks, a few hundred dollars in Claude tokens.
Read →I led an evaluation of 14 EHRs before we chose one. The number that matters most isn't the licensing fee — it's the engineering tax. What a real scorecard looks like, and what a 12-clinic cutover in a single weekend taught us about the cost of a bad architecture decision.
Read →I built a clinical-facing product in weeks, solo, with Claude Code — technically ready to ship. Then I was asked to hold, not for a bug, but to prepare the humans. AI has widened the gap between technically ready and organizationally ready.
Read →The most diagnostic question I ask in any discovery conversation. Every instance of copy-paste points at an integration decision that was never made — and AI scribes are exposing that debt faster than teams can budget for it.
Read →Faster patching lowers the signal on a platform problem that still needs a real decision. Same categories of bugs, faster fixes — that's not an engineering problem. That's a platform problem.
Read →A Nature paper introduced MIRA — an autonomous agent operating within EHR environments that outperformed experienced physicians 87.8% to 78.1%. What it means for product builders in healthcare, and the questions it raises about trust, governance, and responsible deployment.
Read →Notes from a panel on lean development in the vibecode era for healthtech: what AI is and isn't ready for, why build/buy/partner is now a remix, and what's actually proprietary when code is cheap to produce.
Read →After five and a half years — from one clinic in Brentwood to 20+ clinics, 100+ providers, and a Series B platform — I'm handing over the baton and carving my own path. Why I'm launching ARK Foundry, and the work I'm most proud of.
Read →AI tools can extend what a product manager can build, but the ceiling on that work isn't the tools — it's the domain depth, data fluency, and self-awareness to know where your lane ends. Three things I brought to the table before I wrote a single line of code.
Read →AI has collapsed the timeline for shipping. Our organizational processes haven't kept up. When a feature ships in hours, the bottleneck is no longer building — it's adoption.
Read →A colleague built an OCR feature in three minutes. I ordered groceries from 30,000 feet and they beat me home. Some thoughts on what it feels like to live through this moment — and why the best of it never announces itself.
Read →