Automatic screening of photos and documents — metadata forensics, tamper detection and cross-checks against the weather that really happened. Honest files sail through; suspicious ones get flagged with reasons.
Each photo and document is screened for what it says about itself, whether it has been altered — and whether it matches reality at the claimed time and place.
Capture time, device, GPS and edit history read from every file and compared with the claim.
Signs of editing, splicing or AI-generated imagery flagged with confidence scores.
Does the damage match the weather that actually occurred there? Our 20-year archive answers.
Your rules turn scores into statuses — pass, review or reject — inside the claims flow.
Flagged files arrive with human-readable explanations, not black-box scores.
Screen single files or whole backlogs; results as JSON with full detail.
Photos and documents arrive via API straight from your claims system — or in bulk for a backlog audit.
Metadata, manipulation signals and the weather cross-check run automatically on each file.
Clean files pass silently; suspicious ones land in a review queue with the evidence attached.
A queue of flagged files with side-by-side evidence — built for fast decisions.
JSON in, verdict out — wired into intake so screening happens before an adjuster ever looks.
Every flag comes with a confidence score and reasons, and thresholds are tunable — you decide what auto-passes and what a human sees. A pilot on your closed cases shows the real numbers.
We flag statistical signs of generated or heavily edited imagery. It’s an arms race, which is why we layer it with metadata and the weather cross-check — context is much harder to fake than pixels.
Missing or stripped metadata is itself a signal, and the weather cross-check still works from the claimed time and place.
Screening runs at intake via API, so results are attached before triage. The dashboard is optional for the review queue.
“Fraud rarely survives context. A photo carries its own testimony — when it was taken, where, and whether the sky agrees with the story.”
Send a handful of settled claims — including the ones that turned out to be fraud — and we’ll show you what the screening would have flagged, and why.
See a screening demo