Insurance · Case study

Insurance major in Europe

Manual claims processing averaged 14 days. Customer NPS was a chronic underperformer for the brand.

The context

A major European insurer's claims process averaged fourteen days of manual handling, and customer NPS suffered for it. The bottleneck was triage: every claim was read, classified and routed by hand against a dense policy schedule.

At a glance

Sector
Insurance
Headline result
<48hr

What we did

The approach.

  1. 01

    Built triage agents

    AI claims-triage agents on private-cloud LLMs read each claim and classified it against the policy schedule, with human-in-the-loop review on anything ambiguous.

  2. 02

    Grounded every decision

    Each triage decision is grounded in and cites the specific policy clauses behind it, so the audit trail holds up against the schedule and the regulator.

  3. 03

    Kept humans in control

    Adjusters review and override at the edges; the agents handle the routine volume that was previously eating the team's week.

  4. 04

    Handed back ownership

    The insurer operates the agents and the eval harness in their own cloud, with full auditability and no third-party data exposure.

The outcome

What changed.

<48hr

claims processing time, down from 14+ days

Full

auditability against the policy schedule

↑ NPS

from a chronic brand underperformer

Let's talk

Want an outcome like this?

A senior engineer on the call from day one. Two weeks to a written architecture and a working reference build.