15+ years building every layer of healthcare: the clinical stack, AI in regulated practice, and the payer economics underneath.
Run as repeatable playbooks, priced by outcome, not hours.
A 30-minute call to get a first read on the layer you’re building — and a proposal for a 2–4 week problem-framing engagement.
Most healthcare builds are not new problems. I have shipped each layer before, as the executive accountable for the outcome.
An EHR that fits the practice, records that arrive before the visit, prescribing that works, billing that gets paid, AI deployed where care actually happens — and the plan and payer economics that make it hold. I have been the product and technology executive accountable for each of these, across payer and provider.
EHR selection or a custom build prototyped with AI in days (e.g. Medplum, Elation, custom), clinical operations and credentialing, records and interoperability (Metriport, Zus), prescribing, prior auth and labs (Photon, DoseSpot, Surescripts, Junction, Quest) — the workflow built around the clinic, not the vendor.
Record review, ambient scribing, summaries, inbox and call deflection — before, during, and after the visit. Deployed in regulated settings with evals, guardrails, and PHI segmentation: separated, compliant, auditable, and a human gate on every step.
Billing, claims, RCM and TPA operations, value-based-care readiness, and plan & benefits design — self-funded and level-funded, single and multi-payer, contracted and shipped, with unit economics that hold.
Each is a standard project with a known shape, run before, priced by playbook rather than hours, with the companies where I shipped it.
Onboarding and activation, async and sync care models, retention and engagement.
Medplum, Elation, or custom; credentialing, provider data & directories; workflows built around the clinic.
Record review, scribing, summaries, inbox and call deflection — Anthropic, AWS, or third party (Nabla, Zoom).
Retrieval and aggregation via Metriport or Zus, normalized, inside the clinical workflow.
E-prescribing via Photon, DoseSpot, Surescripts; pharmacy routing; labs via Junction, Quest.
Custom and insurance billing, RCM, claims adjudication in house and third party.
GLP-1 and weight management, coaching, dietitian services, with unit economics that hold.
Engagement at payer scale, measurement, contracting readiness.
Self-funded and level-funded, single and multi-payer, tiered networks, plan design.
Not a deck. AI running in clinical care, a custom EMR prototyped in days, insurance and TPA economics, and virtual care at national scale.
As SVP Product & Technology (CPTO) at Form Health (obesity and cardiometabolic care): put AI directly into the clinical workflow — ~50% less clinical visit time, patient access 10× faster, ~50% of inbound calls AI-handled — while scaling ARR and patients 15× and doubling gross margin above benchmark.
Clinical programs were running on slow homegrown software. I prototyped a path forward in days with AI and short iterations with the clinical and operations team, built on the Medplum headless open-source platform.
As VP Product & Engineering at Sana Benefits (self-funded insurance for SMBs): grew 50%+ while doubling gross margin above benchmark and cutting R&D ~40%; managed benefits and enrollment for 2,000+ businesses, 30,000+ members, and launched Sana Care, a new virtual primary-care line.
At a Fortune-10 health plan: defined the virtual-care strategy, launched UHC’s Virtual-First Health Plan (NavigateNOW), and shipped an end-to-end Virtual Care Hub to ~40M members across Optum Care and partners, on a member experience at ~4M monthly active users.
Early product hire at a benefits-administration and TPA platform: 0→USD 40M+ ARR on USD 1B+/yr in healthcare spend, with a model independently linked to 45% lower employer total cost (JAMA Network Open).
A concept for how it could work across every system, workflow, and role: connecting your clinical systems and the evidence you trust, with the clinician always in control, live in minutes.
Most healthcare builds are not new problems. They are known projects with known shapes: an EHR that fits the practice, records that arrive before the visit, prescribing that works, billing that gets paid. I have run each of these before, as the product and technology executive accountable for the outcome.
So I don’t start with the vendor or the demo. Each playbook is a repeatable engagement with fixed scope, a clear definition of done, and a price by playbook, not hours — and AI enters where it earns its place, gated by a human, safe to run under regulation.
More learnings from my decades in tech and digital health on Substack →
Most engagements start with a 2–4 week problem-framing sprint, priced by role and cadence, not hours. Then one playbook ships to done in 2–3 months, or several in sequence under one roadmap.
Start from the layer that actually matters — the EHR, the records gap, the AI opportunity, the plan economics — and define done up front.
Run the known shape with a known team: fixed scope, compliance path defined, AI where it earns its place, a human gate on every step.
One playbook, then the next, under one roadmap — and your team running each this way after I’m gone.
See the full set of capabilities →