SCOR

As a global reinsurer operating across life, health, and property & casualty lines, SCOR's underwriting and claims teams routinely process vast volumes of incoming documents — medical records, policy submissions, claims files — spanning multiple languages, formats, and complexity levels, including handwritten materials. Manual review of these documents created significant bottlenecks in risk assessment workflows, slowing the speed at which underwriters could evaluate submissions and render decisions. Inconsistencies introduced by human fatigue and varied expertise across a global workforce further complicated quality control. In a business where accurate and timely risk decisions directly affect profitability, these inefficiencies represented a material operational liability.

Under its Forward 2026 strategic plan, SCOR developed AI Assistant — a proprietary, cloud-based Generative AI tool built internally by and for underwriters. Rather than relying on a third-party vendor, SCOR engineered the system to extract structured data from unstructured documents regardless of language, length, format, or the presence of handwritten text. The rollout followed a disciplined pilot-first approach: select medical underwriters in the Life & Health division tested the system in 2024, providing feedback that shaped refinement before broader deployment. The tool was subsequently scaled to more than 150 underwriting and claims professionals and now ingests approximately one million pages per month. Following internal validation, SCOR commercialized AI Assistant through its SCOR Digital Solutions subsidiary, making it available to reinsurance clients.

The headline outcome is a 30% reduction in time spent on medical underwriting within the Life & Health division — directly accelerating risk assessment cycles. At scale, the system processes roughly one million pages per month with approximately 90% accuracy across key data fields, replacing error-prone manual data entry. Beyond speed and accuracy:

• Consistency of underwriting and claims decisions has improved as a shared AI layer standardizes how information is extracted and surfaced
• Adoption reached 150+ underwriting and claims experts, indicating strong operational embedding
• The internal tool has been productized, generating a new commercial revenue stream via SCOR Digital Solutions

• Build for internal use first: SCOR's decision to design AI Assistant specifically for underwriters — not as a generic enterprise tool — drove both adoption and measurable outcomes before any commercial offering.
• Pilot with domain experts: Starting with select medical underwriters in 2024 enabled targeted feedback and reduced rollout risk at scale.
• Keep humans accountable for risk decisions: AI handles extraction; underwriters retain final judgment — a critical governance boundary in regulated reinsurance environments.
• Proprietary systems can become products: Internally validated AI tools built on real operational data carry credibility that third-party solutions often lack, creating a path to commercialization.