$26M · Metriport + 3 more

What they're building

Metriport is open-source infrastructure for the one thing American healthcare has never solved: putting a complete picture of a patient in front of the clinician treating them right now. The company maintains dozens of connections into the national health information networks, hospitals, pharmacies and clinics, then runs the unglamorous machinery on top of them. Matching, extraction, standardization, deduplication, enrichment. A pile of contradictory documents from twelve systems becomes one longitudinal record. The delivery surface is deliberately plural. The same unified record ships as a FHIR-native API, as an analytics-ready data warehouse schema, as pre-built apps that run natively inside major EHRs, and as embeddable dashboard and iframe components for teams that want a UI rather than a pipeline. On top sits the analytics layer: medical record summarization, condition suspecting, care gap identification, and real-time alerts when a patient moves through the system. The concrete version of the pitch is the Sollis Health case in the funding announcement. A patient presents with abdominal pain and cannot recall the details of a prior endoscopy. Metriport surfaces the earlier operative report, the ulcer gets identified, and nobody orders redundant imaging. That is the whole product in one story: not a model that reasons better, but a retrieval problem that was never solved. Open source here is strategy, not marketing. Record matching and deduplication are exactly the kind of adversarial, edge-case-ridden work where a closed vendor accretes silent errors and customers have no way to audit them. Metriport also runs an in-house legal team that vets which organizations get to reach health information networks through its API, which is an unusual thing for an infrastructure startup to build early and tells you what they think the real risk is.

Why this matters

Every clinical AI company in this cycle is building on top of patient data it does not actually have. The models are commoditizing fast; the connections, the matching logic, and the network access agreements are not. Metriport's co-founder Colin Elsinga put the thesis plainly in the announcement: usable data from longitudinal patient records is the foundation that all downstream processes and technologies, including AI, are built on. That is a claim about where the durable margin sits. The customer list is the evidence that matters more than any market-size number. Amazon One Medical, Sollis Health and Color Health are organizations with real compliance departments and real alternatives, and they picked an open-source startup with under $3M raised before this round. The company also cleared HITRUST r2 certification, which is the gate that stops most early infrastructure vendors from selling into health systems at all. The timing argument is that going from $2.4M to $28.4M total raised is not an incremental step. It is a company that spent three years building distribution-grade plumbing quietly and has now been handed the money to staff it properly, all at once.