Overview
Motor claims on the insurance platform: two-stage AI turns accident photos, policy context, and repair data into structured damage insight, cost signals, and risk gauges — Arabic/English for bilingual teams.
the insurer platform: AI that supports the decision, speed that supports the customer. In claims, time is money and accuracy is capital — this capability converts accident imagery and policy facts into a clearer view of damage, a defensible cost signal, and fraud/driver-risk indicators in minutes instead of days.
Inputs stay familiar: what adjusters already capture in the claim file and policy — accident date, location, description, severity, weather/road, injuries, third parties, police report when present; vehicle and coverage details; customer and driver history; photos (including policy baseline shots matched by angle to separate new damage from prior damage); injuries and medical estimates; extra costs and recoveries; approved parts tables with reference pricing; and contextual financial cues (agent, premium status, bounced cheques when available). The model is not guessing in a vacuum — it runs on a complete claim context.
Two complementary stages: (1) Per-image vision — each accident photo is analysed (in parallel when needed) with its matching policy baseline when linked; output includes bilingual damage narrative, structured damage list, severity, confidence, suspicious cues, and prior-damage flags. (2) Whole-file reasoning — all image findings are merged with the rest of the dossier for holistic consistency: vehicle match, environment, pricing linkage, injury plausibility, total cost estimate, fraud/driver-risk indicators, and an executive summary for specialists.
On screen: a structured report with gauges (confidence, fraud risk, driver risk), decision summary, token/time metadata, and traceable sections so reviewers can move from registration to assessment with less repetitive manual work — while final underwriting decisions remain human-led where policy requires.