RAD Medical LLC
The experienced clinician in the room.
My name is Alex Butler. I'm a pediatric hospitalist. Early-stage health tech teams bring me in to tell them whether the thing they're building will survive contact with a real clinic.
Start a conversation How I work
The problem I solve
A lot of clinical software works in the demo, but struggles in the clinic or the hospital.
That's usually nobody's fault. It comes down to a few dozen small decisions, made quietly and early, by people who have never carried a pager. I've made some of those decisions myself and gotten them wrong. The job is to be in the room while they're still cheap to change.
What I bring
I still see patients
When I tell you how a clinician will react to your product, I'm describing shifts I worked this month. I round, I take call, I use the software everyone complains about.
I know the systems
A master's in clinical informatics means I can sit in the schema conversation, not just the medical one. When it gets deeper than that, Derrick Chu, who works with me, is an MD-PhD who builds in Epic.
I've shipped product too
I have worked as Chief Product Officer at two healthcare start-ups. My fractional work runs from seed companies with a demo to Series C companies with live deployments inside health systems. I can handle communication between providers and your dev team.
From the room
What this looks like in practice.
A team was two quarters into planning a standalone provider portal, but nobody had asked the providers. They already have an inbox they hate and a login they resent, and a twelfth one was not going to win. I helped clear the portal from the roadmap in the second week, and the work refocused where providers needed it more.
A triage tool had to decide what to do when a referral arrived missing a piece of information. The engineering instinct was to treat missing as negative and keep the queue moving, which is reasonable, but wrong: missing does not mean absent. It means somebody has to go look, and the patients whose charts are thinnest are usually the ones with the least capacity to advocate for themselves. The rule became: unknown gets verified.
An AI-drafted note scored well on accuracy, but the clinicians still hated it. It was correct in somebody else's voice. This isn't an engineering problem but an evaluation oversight and the fix was to collect each clinician's own notes up front and measure edits per note instead of accuracy alone.
Who this is for
Seed to Series B, with a clinical surface and no clinician on the team.
- Founders building something that touches real clinical decisions, who'd rather hear what a doctor thinks now than after launch.
- Investors who want a clinician's read on a company before the money goes out.
- Teams whose product has started making clinical calls faster than anyone in-house can check them. That's the stage where the architecture hardens around whatever you assumed, and where a fractional Chief Medical Officer earns more than a full-time hire would.
How I think
I write about where clinical judgment meets what gets built.
A few questions keep pulling at me: when a human really needs to be in the loop, and when "human in the loop" is just there for the lawyers. Who's accountable when a system does exactly what it was built to do and the patient is still worse off. What happens to patients as AI works its way into the visit.
Building something clinical?
Clinical problems get expensive after launch. Before is cheap.