Healthcare is the largest industry tag in the catalog — 2,017 companies claim it, and 99.4% of them claim it as part of their own identity rather than having it applied to them. It branches into nine narrower tags, from Electronic Health Records to Clinical Trials. And it is the weakest large cohort we score. The average Kin Score across those providers is 15.8, against 22.7 for the rest of the catalog. The median is 10.7 against 16.3. Half of healthcare — 1,008 companies — sits in the minimal band, meaning almost no public developer surface at all, against 34.7% of everyone else. This is an industry that runs on interoperability standards and publishes less machine-readable API surface than the market it is a part of.
The top of the tag does not look like that at all. Medplum leads at 87.4, exemplar — an Apache 2.0, FHIR-native developer platform with a FHIR R4 datastore, REST and GraphQL, SMART on FHIR auth and serverless Bots, carried by an operational transparency score of 97.4. Elation Health follows at 83.7, a clinical-first EHR for independent primary care that publishes 126 API pages here and an agentic access profile covering 846 operations, 455 of them acting. Stedi scores 83.1 as an API-first programmable clearinghouse — eligibility checks, claims and remittance advice as JSON over X12 EDI — with perfect marks on access clarity. The Centers for Medicare and Medicaid Services is at 81.3 across 41 API pages of Medicare provider data, quality measures, drug spending and beneficiary claims, which makes the federal payer a better API publisher than almost every company it pays. And drchrono, now part of EverCommerce’s EverHealth portfolio, rounds out the five at 80.9.
Those five average 83.3 — 67 points clear of their own category. That is the real shape of healthcare: a ceiling as high as anyone’s and no middle underneath it. The exemplar share is the giveaway — 1.1% of healthcare providers reach the top band, which is exactly the rest of the catalog’s 1.1%. Healthcare produces world-class API programs at the same rate everyone else does. What it does not produce is the broad middle: developing is 5.6% here against 14.1% outside, and thin is 6.9% against 16.0%. In most categories a company that has not built a great API has at least built a partial one. In healthcare it usually has not built one the public can see, and the cohort falls straight from the exemplars to the floor.
On agent readiness the gap is wider, and it runs in one direction. Healthcare averages 5.3 on Agent Readiness against 11.4 for the rest of the catalog, and 78.5% of these providers are human-only — no machine-readable contract, no documented auth an agent can follow — against 52.6% outside. Of the nineteen dimensions the rubric reads, healthcare is below the rest of the market on seventeen, level on one, and ahead on one by a tenth of a point. The biggest gaps are the most basic things: a machine-readable contract is present for 14.4% of healthcare against 35.8% of everyone else, documented auth for 19.3% against 40.2%, rate-limit signalling for 18.8% against 37.8%. Only 5.3% light the MCP server dimension against 12.4% of the rest of the catalog.
The dimension healthcare should own outright is empty everywhere, including here. Consent and identity — whether a provider documents how an agent acts on behalf of a named person, with consent that can be withdrawn — is lit for 0.3% of healthcare and 0.3% of everyone else. An industry with HIPAA, patient authorization and a federal information-blocking rule has no more documented answer to “who is this agent acting for” than a company selling sneakers. Documented reversibility, which tells an agent how to undo what it just did, is at 2.6% against 7.4% outside. The regulatory machinery is there. The machine-readable expression of it is not, and that is a gap agents will hit long before they hit the EHR integration itself.
One caveat on the cast. Healthcare is not one thing — the catalog splits it into four senses, and the largest is Digital Health and Telehealth at 46.5% of the tag, ahead of drug discovery and genomics at 12.9% and revenue cycle at 8.6%. The five leaders above are systems of record and claims infrastructure; not one of them is a telehealth company. The ranking mixes the senses, so the top of this tag says more about who runs the clinical back office than about who is building the front end of digital health. That gap is worth its own post.
Healthcare’s ceiling is not the problem. Twenty-three companies in this tag score in the exemplar band and several of them are among the best API programs on the network, full stop. The problem is that 1,008 companies in an industry defined by data exchange have published nothing a machine can read, and the standards those companies already implement — FHIR, SMART on FHIR, X12 — are the exact artifacts the Kin Score is looking for. The work is mostly already done internally. Publishing the contract is what separates Medplum and Stedi from the half of this category an agent cannot find the door to.