GenAI claims triage at a top-five UK insurer

The client is a top-five UK motor insurer handling high daily volumes of claims, where the speed and fairness of first notification of loss shape both customer retention and cost. In a competitive personal-lines market, a slow or inconsistent claims experience is felt directly in renewal rates.

Their loss-adjuster team was skilled but expensive and finite, and every routine claim that consumed adjuster time was time not spent on the genuinely complex or high-value files where human judgement pays for itself.

Claims arrived as a jumble of forms, photos, and telematics feeds that a human had to open, read, and route by hand. That manual triage was a bottleneck at exactly the moment the customer was most anxious for movement.

The volume made consistency hard: two similar claims could take different routes depending on who picked them up, and nothing produced the clean, per-decision audit trail the loss-adjuster and complaints teams needed to stand behind an outcome.

The insurer wanted to fast-track the routine majority without weakening control or letting leakage through. Speeding up settlement and tightening oversight usually pull against each other, and they needed both at once.

We built a multi-modal triage engine that reads every document, image, and telematics feed on arrival, classifies the claim, and routes it, pricing the straightforward cases straight through while flagging anything ambiguous for a human.

Every decision carries a full audit trail, so the loss-adjuster team can reconstruct why a claim went where it did. Low-confidence and high-value files still route to an adjuster, keeping judgement where it belongs and the automation trustworthy.

We trained on the insurer's historical claims and shadow-tested against real adjuster decisions before going live, so the engine matched their people's judgement before it was allowed to act on volume.