Zurich Insurance Group

Zurich Cover-More, Zurich Insurance Group's global travel and assistance subsidiary, fields more than one million customer calls annually — a volume that placed enormous pressure on its claims operation. The core bottleneck was document handling: adjusters had to manually review PDFs, photographs, and handwritten submissions before any coverage decision could be made. This sequential, labor-intensive process stretched processing times to hours, creating friction for policyholders at exactly the moment they needed fast resolution. In Chile, the same structural problem afflicted the healthcare claims book, where slow cycle times drove up unit costs and suppressed renewal rates — a direct threat to portfolio retention in a competitive market.

Zurich addressed the travel claims problem through CLARA, a multi-agent generative AI system built in partnership with AgentricAI. CLARA ingests unstructured inputs — PDFs, photos, handwritten notes — and orchestrates a pipeline that verifies coverage eligibility, renders an approval or denial decision, drafts the customer-facing email, and calculates the payout amount. Critically, the system operates under continuous human supervision, preserving the oversight layer required for regulatory compliance in P&C. The deployment followed a controlled pilot before a planned production rollout across Australia and the United States. In Chile, Zurich partnered with InsurTech LISA to deploy a separate no-code claims processing engine purpose-built for healthcare, enabling rapid adoption without deep engineering resources on the insurer's side.

CLARA compressed travel claims processing from hours to an average of 47 seconds, with a ceiling under two minutes. The Chile healthcare deployment produced a distinct but equally significant set of outcomes:

• ~67% of claims now processed without human intervention
• Settlement achieved in under 10 minutes, end-to-end
• Unit costs reduced by approximately one-third
• Annual savings of nearly $1 million in the Chilean book alone
• Health insurance renewal rates exceeded 90%, reaching record levels

Production expansion of the travel claims system to Australia and the US was planned for end-of-year, indicating confidence in the pilot's stability and scalability.

• Multi-agent architectures that decompose claims into discrete AI-handled steps — ingestion, verification, decision, communication, calculation — can achieve sub-minute cycle times while preserving required human oversight.
• No-code platforms lower the implementation barrier for insurers without large engineering teams, enabling meaningful automation at scale without bespoke infrastructure.
• Speed improvements in claims settlement have a measurable downstream effect on retention; the Chile data linking sub-ten-minute payouts to 90%+ renewal rates illustrates this directly.
• Piloting in a single market before broader rollout allows teams to validate model behavior and compliance posture before committing to enterprise-wide production deployment.
• Pairing AI efficiency with customer-visible speed — not just internal cost reduction — is what drives retention lift.