Income and asset fraud starts with a document — an edited pay stub, a fabricated bank statement, an invoice with a bumped total. Docurensic screens each one for the fingerprints of tampering before capital goes out the door.
A constructed example — not a real customer file — showing the kind of finding Docurensic surfaces on a bank statement.
Opening plus deposits minus withdrawals doesn't equal the stated closing figure, and that figure carries a font and baseline mismatch against the column above it.
Forensics · arithmetic + font · strongA later revision touches only the balance region, timestamped after the statement's own period.
Metadata · revision · indicativeUnderwriting is only as good as the documents it trusts. A number changed in a PDF viewer looks identical to the real thing on screen — but the file remembers. Revision markers, metadata timestamps, and font substitutions expose edits the eye can't see.
The reasoning and validation layers add a second dimension: do the totals reconcile, do the dates make sense, does the story the document tells actually hold together?
Five layers, one verdict. Every document runs through security, forensics, reasoning, validation, and external lookups — see how the engine works.
Document-specific guides: spotting a fake bank statement · spotting a fake pay stub.
Yes. It checks the file's structure and metadata for signs of editing, validates that balances and transactions reconcile, and flags statements that were generated or altered rather than issued by a bank.
The AI/content-trust layer is built for exactly this — synthetic documents and AI-generated text and images that increasingly show up in loan files.
You choose. Store keeps an encrypted copy in your File Vault; Storageless creates no new copy at all. Full detail is on our Security page.
More industries
Freight & logistics · Insurance · HR & onboarding · Legal · All use cases
Free to start. No install, no sales call.