Doctored payslips and fabricated bank statements are trivial to produce and hard to catch by eye. Seal income and identity evidence at the moment it is captured, and the question stops being whether the document looks right.
A payslip is a PDF. Changing a number takes seconds and leaves no visible trace. Screenshots of banking apps are worse: they carry no provenance at all and are indistinguishable from a mock-up.
Fabricated income does not fail at the document check. It fails eighteen months later as an arrears case, by which time the loan is on book and the loss is real.
Reading every document by eye is expensive, slow, and no better than chance against a competent forgery. Speed and diligence pull against each other.
The borrower photographs the payslip or ID through your app. CapSeal measures whether a real lens saw a real document - parallax against device motion, depth relief, screen-replay detection - and binds the result into the file.
Your origination system posts the file and receives a verdict with a basis. Sealed captures come back proven; anything else comes back with the evidence you would need to decide.
Proven and declared results route automatically. Everything else goes to a human with the full report attached, so the reviewer starts informed rather than cold.
CapSeal does not replace your origination process - it adds two touchpoints to the one you already run. Documents are sealed at upload, a verdict attaches at assessment, and each file routes down one of three paths. Every outcome lands back in your LOS with the proof stored on the loan file.
Note the middle path. Paper and branch applications route to your normal document review exactly as they do today - never auto-declined, and no borrower is worse off for not having used the app.
CapSeal establishes that a document was captured by a real camera pointed at a real thing, and has not changed since.
The full position is in the Product Disclosure. We would rather you knew the boundary before you bought than after.
A bottom-up model of fabricated-document exposure for a digital lender writing 40,000 applications a year, built on published research from Point Predictive and industry survey data. Six pages, every assumption on the page.
The honest boundary, stated in the brief itself: the model covers fabricated-income and identity documents, the category capture-time provenance directly prevents, and uses a 3% fabrication rate well below the published 1-in-10 paystub figure. It does not claim staged-borrower rings, genuine documents describing an untrue situation, or fraud that never touches a document. The figures are modelled from published ranges, not results from a named lender.
Try the verification half right now, with no signup, at capseal.ai/check.
A pilot takes one line of your origination flow and one API call.
Book a pilot Check a file first Lending brief (PDF) The lender case, in full