Insurance, Motor Claims
The first insurer in India to trust AI with a claim
In June 2018, India's second-largest private general insurer did something no competitor was willing to do: it let a machine look at a damaged vehicle and decide what happened next. Five and a half years later, that decision is still running in production.
Jun 10, 2018
3 min read
The cost that scaled with every single claim
Bajaj Allianz General Insurance Company (BAGIC) processes motor claims at national scale. Every one of them, historically, required the same three things — a surveyor, a physical location, and a trip. That model has a hard ceiling. Inspection cost does not fall as volume rises; it rises with it. And the assessment at the end of the trip is a human judgement call on a borderline panel, made under time pressure, at a rate that varies by inspector and by hour of the day.The objective was never speed. It was cost per inspection.
BAGIC was explicit about it. Reduce what it costs to inspect a vehicle, without reducing the quality of the assessment that comes out the other end.The approach: put the insurer at the centre, not the technology
CamCom ran a proof of concept in June 2018 to demonstrate what computer vision could do inside a live surveyor workflow. The client moved to a full trial between December 2018 and January 2019. The design principle was deliberate. Rather than replacing the surveyor’s process, CamCom built around it — a contactless inspection platform intuitive enough to sit inside an existing, human-centred workflow. Photographs of the vehicle go in. A remote assessment of external damage comes out.The solution: damage assessment inside BAGIC’s own app
CamCom built an AI-based remote automobile inspection system and delivered it inside the mobile app BAGIC had already deployed to its surveyors. No new tool to learn. No parallel system to maintain.- Guided capture: On-screen stencils and in-app instructions direct the user to the right angle, distance and framing for every panel — the single largest determinant of assessment accuracy.
- Automated damage detection: Advanced deep learning methods assess damage across the vehicle’s external panels without human input.
- Severity and decision output: The system returns the intensity of each detected damage and a repair- versus-replace decision on the part.
- Straight into settlement: That output feeds the claim, settling it in a fraction of the time the same claim took manually.
The impact
The trials were a success and the system went into production in February 2019.- Business continuity, guaranteed: Human intervention was eliminated from the assessment step, so the process no longer depends on surveyor availability or physical access.
- Regulatory compliance at speed: Claims under INR 50,000 now settle within the timelines IRDAI requires.
- A permanent visual audit trail: Every inspection contributes to a visual database of vehicles with a dated assessment of condition — evidence that holds up long after the claim closes.
