Solutions

Start with motor.
Expand across the book.

CapSeal lands on the line where photo-based fraud is loudest, then extends wherever claims rely on images or documents. The mechanism is the same everywhere - trust the evidence at capture - but the fraud pattern it kills differs by line.

By line of business

Where it fits first, and next

START HERE

Motor

Highest photo volume and highest fraud - the natural beachhead. Kills edited and AI-generated damage photos, re-used images from prior claims, and staged-damage submissions.

  • Damage photos sealed at first notice of loss
  • Reused-image detection across claims
  • Instant settlement for attested minor claims
EXPAND

Property & home

Fabricated or exaggerated damage imagery and doctored contents inventories. Provenance-sealed capture makes "before/after" manipulation and stock-image submission unworkable.

  • Sealed property and contents photos
  • Document provenance for repair estimates
  • Geo and time coherence at capture
EXPAND

Commercial & SME

Higher claim values raise the payoff of fabricated invoices and inflated estimates. Document sealing and template-lineage checks target manipulated repair and replacement paperwork.

  • Sealed invoices and assessment reports
  • Tamper-evident supporting documents
  • Cross-carrier verification on large losses
WITH CARE

Health & injury

Synthetic medical certificates and fabricated reports are a fast-growing vector. CapSeal seals document provenance - but note the honest limit below: it proves the document is real, not that the underlying claim is truthful.

  • Provenance on medical and supporting documents
  • Detection of AI-generated certificate templates
  • Feeds the intent layer for deeper review
The honest boundary: CapSeal proves evidence is real - real device, real moment, unaltered. It does not read intent. For lines where the fraud is about what happened rather than whether the image was faked, CapSeal is the intake layer that feeds the intelligence engine, not the whole answer. We tell buyers exactly what each line can and can't promise.
What to expect

A land-and-expand path

PHASE 1

Pilot on motor

One line, 60-90 days, measured on auto-pay rate and fraud caught at intake.

PHASE 2

Roll out the line

Full production on motor, tuned policy, dashboard live for your SIU team.

PHASE 3

Add adjacent lines

Property and commercial, reusing the same SDK and verdict engine.

PHASE 4

Join the network

Cross-carrier verification switches on the double-claim moat.

Inside your claims workflow

Where CapSeal sits in the claim lifecycle

CapSeal does not replace your claims process - it adds two touchpoints to the one you already run. Evidence is sealed at lodgement, a verdict attaches at intake, and triage routes each claim down one of three paths. Every outcome lands back in your core system with the proof stored on the claim file.

YOUR PROCESS • STAGE 1 Claim lodged (FNOL) Customer photographs damage in your existing claims app CAPSEAL: evidence sealed at capture YOUR PROCESS • STAGE 2 Intake & triage Claim registered in your core system as normal CAPSEAL: verdict attached to claim ATTESTED-GENUINE Auto-pay Straight-through settlement in minutes UNVERIFIED Normal review Your existing queue, unchanged. Never auto-rejected. TAMPERED / SYNTHETIC Held for SIU Payment stopped; investigator gets the claim with proof attached YOUR PROCESS • STAGE 3 Settlement & audit Every outcome lands back in your core system. The proof object is stored with the claim file - replayable for audit, dispute, regulator or court, years later, without calling CapSeal. INTEGRATION FOOTPRINT Runs beside Guidewire, Duck Creek or your native core - an SDK at capture and one API call at intake. No core-system migration. Claims not captured in-app simply route as unverified.

Note the middle path. A claim not captured in-app routes to your normal review queue exactly as it does today - it is never auto-rejected, and the customer is never worse off for having an old handset or emailing a photograph in.

Industry brief

The motor numbers, with every input cited

A bottom-up model of manipulated-image leakage for a carrier writing 200,000 motor claims a year, built on published 2026 research from Verisk, Shift Technology, Admiral and the Coalition Against Insurance Fraud. Seven pages, every assumption on the page.

$8.0M
net-new leakage prevented per year, base case
20-30%
of claims now include AI-altered media
~75%
of flagged claims never investigated

The honest boundary, stated in the brief itself: this model covers manipulated-image fraud, the category capture-time provenance directly prevents. It does not claim staged-accident rings, padded but genuine estimates, or claims where the photo is real and the story is not. The figures are modelled from published ranges, not results from a named carrier.

Motor brief (PDF)

Pick your loudest line. We'll prove it there.

Motor is the usual start, but we'll run a pilot wherever your evidence fraud is worst.

Request a pilot Model the ROI