AWS and Strands demonstrated an agentic architecture that automates first notice of loss (FNOL) intake by combining the Strands Agents SDK for domain reasoning with Amazon Bedrock AgentCore Browser Tool for live portal interaction (AWS Machine Learning Blog). The system processes multimodal evidence—photos, videos, scanned documents, and recorded notes—at ingestion time, applying insurance-specific business rules to interpret, validate, and correlate artifacts before human review. Amazon Nova Act drives browser-level interactions while Strands Agents apply expert logic in the background, separating mechanical screen work from domain reasoning.
For commerce and claims practitioners, this approach addresses a critical bottleneck: manual FNOL processing consumes substantial expert time on repetitive intake validation that delays claim cycle time and customer experience (AWS Machine Learning Blog). By automating portal navigation, evidence completeness checks, and artifact interpretation, adjusters receive context-rich, pre-analyzed submissions ready for judgment rather than validation. Consistently tagged evidence becomes a durable operational asset, enabling better routing, pattern analysis, and workflow refinement across the claims lifecycle without modifying existing portals.