In a co-working space south of Market, a digital-health founder pulls up a terminal. Three commands. A fork of medplum/medplum on GitHub. Within an hour, she has a working clinical data repository, a FHIR API, HIPAA-shaped audit logs, and a React component library that can render an intake form before lunch. None of this used to be possible. For twenty years, "building a healthcare app" meant licensing fees with six zeros, sales calls with Epic, and a compliance officer with a furrowed brow.
Medplum is what happened when a small team of engineers - the kind who like reading specs more than reading TechCrunch - decided that the bottleneck wasn't medicine. It was the back end. So they wrote a better one, made it open source, and walked away from the licensing money. The platform they shipped looks, from a distance, like Firebase. Up close, it's something stranger and more useful: a fully fledged headless EHR, FHIR-native to the bone, hosted if you want it, self-hosted if you don't.
That founder with the coffee? She is the product. So is the radiology startup outside Boston using Medplum to wire up an interoperability engine. So is the pediatric clinic in Lagos running it on its own AWS account. The thing about open-source infrastructure is that you stop being able to count your customers. You start counting forks.
Patient data lives in a standards-based store you can query with a modern API. No proprietary schemas. No vendor wedge.
Pre-built UI primitives that render FHIR resources. Skip the boilerplate. Ship the patient portal.
Server-side logic without a server. ePrescribe networks, lab interfaces, HL7 feeds - already wired.
Digital-health startups racing to launch. Hospital innovation teams escaping their legacy stack. Life sciences shops doing remote patient monitoring. Anyone who would rather write code than a check to a legacy vendor.
Apache 2.0 means no vendor lock-in, ever. Pay Medplum if you want the hosted version and support. Don't pay them if you want to run it on your own AWS account. That second sentence is not normal in healthcare.
MIT EECS. Co-founded MedXT (YC W13, acquired by Box). Stints at Facebook and Microsoft. Visiting Group Partner at YC. The product brain.
Director of Engineering at One Medical. Healthcare Engineering Lead at Box. Harvard MBA. The person who actually likes reading the HL7 spec.
Machine learning at Palantir. Customer Impact Lead at Applied Intuition. The operator who turns code into contracts.
A clinical workflow that fits your specialty - DME, pediatrics, behavioral health - without paying a legacy vendor for features you'll never use.
Questionnaires, scheduling, secure messaging, care plans - rendered through React components on top of FHIR.
Bots ingest legacy feeds, normalize them, and push them into the data repository. Radiology, labs, networks - all plugged in.
Structured FHIR resources are training-friendly. Population health dashboards, clinical decision support, MCP servers - layer them on top.
The legal entity is Orangebot, Inc. The product is a FHIR-native developer platform that nobody asked for and many people needed.
Seed funding from the YC batch. Roughly $130k headline number.
The post hits the front page. The framing sticks.
Non-dilutive money for a developer-tools company. Quiet but meaningful.
Healthtech founders default to Medplum. Hospital innovation teams begin self-hosting. The team grows to ~21.
Both videos point at YouTube playlists for Medplum content and founder interviews.
The founder closes her laptop. The mug is empty. On her screen is a clinical workflow that, eighteen months ago, would have required a procurement committee, a vendor demo, and a contract with a multi-year escape clause. Instead, she has a fork. She has a license that lets her keep the fork. She has a hosted option if she wants someone else to handle the AWS bill, and a self-host option if she'd rather be the one paged at 3 a.m.
This is the strange, quiet thing about Medplum. It does not advertise itself as a revolution. It is too engineering-poker-faced for that. It just removes a step from a long list of steps - the step where you had to choose between owning your infrastructure and owning your time. Take that step away, and a lot of healthcare apps that previously couldn't exist begin to exist.
The coffee cup goes in the dishwasher. The patient portal goes into staging. Somewhere in a server, a FHIR resource is created. The plumbing works. The founder gets to think about something else.
Which was the point all along.