Zurich UK pilots generative AI to identify claims trends and enable proactive risk mitigation

Before generative AI, it was very challenging for Zurich to train models to process the various forms of unstructured data received around claims. Claims handlers struggled to receive the right documents at the right times, and underwriters frequently had to manually correct and 'fix' data before making decisions, slowing the process and contributing to burnout.

Zurich UK is piloting generative AI to ingest and analyze unstructured claims documents, identify emerging trends, and surface relevant documents to claims handlers and underwriters automatically. The approach follows a 'build once, use many' philosophy where AI tools built within the group are reused across the entire Zurich group.

Claims handlers report being significantly happier receiving the right documents at the right times and are processing work faster. Underwriters no longer need to manually fix data before making decisions. The generative AI data ingestion tools are already live across the business, with success measured primarily through positive feedback from business users.

• A 'build once, use many' group-wide AI strategy amplifies the value of each individual AI tool by enabling reuse across the entire organization.
• Generative AI unlocks unstructured data processing that was previously impractical to automate, dramatically improving data quality for downstream decision-makers.
• Proactive trend identification in claims data may enable insurers to help customers mitigate risk before problems occur, shifting from reactive to anticipatory claims management.