AI damage detection from the photos.
A single AI vision call against every photo on the lot. Per-panel severity, structural red flags, and an honest list of the angles the listing didn't show you.
Per-panel damage scoring
Every panel that matters to a salvage repair gets a severity tag and a one-line note. The AI cites the photo it's looking at, so you can scroll back to the original.
- Front bumper · light · moderate · severe · replace
- Bonnet · light · moderate · severe · replace
- Wing L / R · per side
- Quarter L / R · per side
- Rear bumper · severity + note
- Roof · dent / crease / hail
- Doors L / R · per side
- Glass and wheels · per corner
The eleven signals.
Beyond panel-level severity, the vision pass scans for the structural and red-flag patterns that decide whether a lot is repairable at all.
Deployed bags add SRS module, seatbelt pretensioners, and clock-spring replacement to any cosmetic repair.
Suspected unibody or chassis-rail damage. Frame-jig work shifts the economics.
Buckled core support — a tell for high-energy impacts that bend longitudinals.
B-pillar intrusion or door-frame distortion. Typically uneconomic to repair properly.
Water lines on door cards, dash, or carpet. Electronics rot follows weeks later.
Burned wiring loom, melted plastics, scorched paint. Wire-by-wire diagnostic territory.
Fluid under the lot in the photos — head-gasket, sump, or transmission seal worth investigating.
Prior repair flagged. Tells you whether you're the second person to fix this car.
Poor prior bodywork. Often hides worse than what's been declared.
Mismatched panels, weld lines in unusual places. Walk away territory.
Catalytic converter, headlights, ECU, infotainment. Adds parts cost or signals theft recovery.
Once damage is scored, the repair plan turns it into line items — and the risk flags catch the patterns this page doesn't cover.
Interior, wheels, glass.
A 1–5 interior score with explicit flags: seat tears, dash damage, headliner staining, missing infotainment, blown airbag covers, water or smoke staining. Plus a per-corner triage on rim curb-rash, tyre tread, and missing wheels — and a glance at the windscreen, side-glass, and sunroof.
Photo-completeness audit.
The most useful thing AI can do for a salvage buyer is admit what it can't see. Every analysis surfaces the missing angles — "no engine bay photo", "no underside", "no VIN plate" — so you know where to send your inspector or how much to discount your max bid for risk you can't price.
Confidence is honest. Photo-completeness gaps feed directly into the analysis confidence band. A lot with no engine-bay photo can't get a high-confidence "engine starts" call from photos alone — the chat will tell you so.
How it stays cheap.
One vision call covers the panel scores, signals, interior, wheels, glass, and photo audit — so you're not paying per analysis-line. Photos travel through to the AI in memory only and are discarded; nothing is persisted as image bytes anywhere in the system.