Not a number you trust. An argument you can defend.
Every price Vlada returns traces to the origin file it came from and can be replayed to the same result. Fiduciaries can't cite a black box — so we don't ship one.
The five steps, from raw file to defensible price.
Source, Resolve, Price, Prove, Compare. Each step adds evidence and carries it forward — nothing downstream can outrun what the file actually says.
Source
We start from the primary files, not a redistributor. Payer machine-readable rate files, hospital standard-charge files, Medicare fee schedules, and CMS provider identity rosters are ingested directly. Each ingest records the file's URL, version, and a content hash of the exact bytes we read.
Resolve
A billing identity is a chain, so we resolve it as one: NPI → TIN → CCN → owner. That turns a raw rate line into a real provider — the individual, the group that bills for them, the facility they practice at, and the corporate parent that owns it — so a price attaches to the party you can actually name.
Price
We read the negotiated rate as written, then place it on a scale: benchmarked to the Medicare allowed amount for the same code and locality, with a fair range for what comparable providers accept. A number becomes a coordinate you can reason about, not a figure floating on its own.
Prove
Every value carries its trace: the origin file, its version, and the content hash of the bytes it came from. The price is replayable — anyone can re-run the exact lookup and land on the same number. That is the difference between a figure you have to trust and one you can defend.
Compare
With identity resolved and prices on a common scale, comparison is honest: across payers, plans, and sites of care, like-for-like. Every row in the comparison keeps its own provenance, so the spread is auditable line by line — not an average that hides its inputs.
Most published rates are noise. We filter them out.
Payer files are enormous, and most of what's in them isn't a real price. We measured it: roughly 92% of raw commercial rate rows are ghost rates — placeholders, unmapped codes, and boilerplate that no one actually pays.
A tool that served those numbers would look comprehensive and be wrong. So the resolution engine computes a quality status for every rate at query time and serves only the ones that clear it. When you ask for a price, you get a rate that resolves to a real provider and survives the filter — never a placeholder dressed up as a quote.
This is a trust feature, not a coverage claim. We would rather return fewer numbers we can stand behind than a bigger count padded with rows that fall apart the moment a fiduciary looks closely.
How strong is the evidence? We label it.
Not every price stands on the same footing, so we say which footing it stands on. Three tiers, and every table carries its provenance columns regardless of tier.
Queryable
The rate exists in a source file, resolves to a real provider, and passes the ghost-rate filter. It answers the question — and it tells you exactly which file it came from.
Certified
The price is corroborated across the source file, the resolved identity chain, and the applicable benchmark. Nothing about the number depends on a name vouching for it.
Data-proven
The number is reproduced against independent evidence — additional source files or an adjudicated claim — so it stands on more than a single publisher's assertion.
Provenance is the product. Security is the floor.
The public pricing data we serve carries no PHI by default — it's hospital and payer files, resolved and benchmarked. When a workflow touches a claims book, it runs counsel-gated, under a BAA, with access scoped to the engagement.
See the trace behind a price.
Bring a code and a plan; we'll show you the number, the file it came from, and the replay that lands on it again.