Healthcare provider data should be traceable to its federal source.

Healthcare provider data is the foundation of every payer credentialing decision, every regulatory compliance check, every patient-finding workflow, and an increasing share of every AI-generated answer about U.S. healthcare. It is also famously stale, inconsistent, and unverifiable. Vendors claim accuracy without showing their sources. State boards publish data that contradicts federal records. AI tools cite numbers that no one can trace back to a primary public-record dataset.

Fonteum exists because the healthcare data category needs the same rigor that financial data got from Bloomberg and Refinitiv in the 1980s — source-traceable, version-controlled, machine-readable, and continuously verifiable. Every provider record we publish carries an eight-tuple provenance contract: source, source URL, dataset ID, snapshot date, methodology version, last-checked timestamp, confidence score, and data availability state. Anyone — a journalist, a researcher, a compliance officer, an AI agent — can re-derive any claim we make from the underlying federal source within minutes.

We publish the methodology, not just the data. Every methodology version bump produces a citeable changelog entry with an RSS feed. Every snapshot we ingest is SHA-256 hashed and verifiable at fonteum.com/verify/[snapshot-id]. The cryptographic chain over those hashes is published at fonteum.com/chain so anyone can confirm the historical record has not been tampered with. The infrastructure that produces the data is the moat; the data itself is just the output.

We publish source-anchored healthcare provider data drawn from six federal public-record sources: CMS NPPES (the National Provider Identifier registry), CMS PECOS (provider enrollment), CMS Care Compare (quality + deficiency reporting across nursing homes, home health, hospice, dialysis, and ambulatory surgery centers), HRSA HPSA (health professional shortage areas), BLS OEWS (occupational employment and wage statistics), and BEA Regional (regional economic accounts). Every record we publish carries the eight-tuple provenance contract above.

We publish research that this data makes possible. Our flagship study on private-equity and REIT ownership in U.S. nursing homes is built on the same source-anchored data layer we license to enterprise customers. The methodology behind every chart, every cited number, and every claim is published alongside the study. Studies are versioned and citeable; methodology updates trigger automatic changelog entries.

We publish the infrastructure other organizations build on top of. Our Model Context Protocol server at mcp.fonteum.com lets any AI agent query the provider graph with provenance returned inline. Our public freshness dashboard at fonteum.com/freshness shows the live status of every source we ingest. Our methodology changelog at fonteum.com/methodology/changelog publishes every version bump with RSS distribution. The webhook registry at fonteum.com/developer/webhooks fires provenance-stamped events on detected provider state changes. None of this is gated; all of it is verifiable.

Editorially, we operate as an independent research organization. We do not accept payment from providers for inclusion, ranking, or favorable framing in any record we publish. We do not allow the providers we cover to influence the methodology that determines how their records are presented. Corrections submitted by providers are evaluated against the federal source-of-record; if the federal record is wrong, we publish the discrepancy and notify the relevant agency.

Operationally, we ship infrastructure that makes our claims verifiable rather than asking readers to trust them. Every snapshot we ingest is hashed; every hash is chained; every chain link is signed; the public key is published on a /.well-known/ path. If any data point we publish does not match what the federal source actually published, anyone can prove it within minutes. This is the functional inverse of how most healthcare data vendors operate, and it is intentional.

Commercially, we license the data layer to enterprise customers (health plans, IDNs, compliance teams) under a contract that requires source-attribution preservation downstream. Customers do not get to repackage our data without the eight-tuple provenance fields intact. The methodology we publish is the contract; we cannot enforce truth claims that we ourselves do not anchor to a federal source.