Every claim
on a document,
on the record.
ProofMart takes a PDF and returns a structured report of findings, each pinned to its pixel, each with the arithmetic. You read the rail, not the doc.
From bytes
to bounded evidence.
Every stage writes back to the same address space: page, x, y, run. When you click a finding in the rail you land on the exact glyph that produced it — because that glyph’s coordinates are what the finding was named after.
Every marker
is a rule with a citation.
One call.
One structured response.
The API is small on purpose. Send a PDF; receive the same finding graph you inspected on the rail. Every field maps back to a rail address.
The JSON is real.
So is the dossier.
The JSON below is what the API returns today. A signed-in scan can also be exported as an Ed25519-signed PDF dossier your reviewers can countersign, built from the same finding graph — anyone can check that signature back against ProofMart, no account required.
Four ways in.
Same engine, every time.
The upload flow above, the public API, webhook delivery, and the CLI all run through the same verification engine — one processing path, not a marketing demo backed by something thinner.
Flat monthly.
No per-document surprises.
One plan, one price, a monthly call limit on /v1/verify. No per-finding or per-scan metering.
For trying ProofMart against real documents before committing to a plan.
For teams running ProofMart in a review or underwriting loop.