AWS has introduced a hands-free first notice of loss (FNOL) system that automates intake validation, reducing adjuster workload by up to 70% during peak claim volumes. The solution combines Strands Agents and Amazon Bedrock AgentCore Browser Tool to streamline the processing of unstructured evidence, such as photos, videos, and scanned documents. This approach enables claims professionals to start with context rather than raw data, improving efficiency and accuracy in the initial stages of claim processing. Source: awsml
The system uses Strands Agents, an open-source SDK, to apply insurance-specific business rules like evidence interpretation and cross-modal correlation. Amazon Bedrock AgentCore Browser Tool provides a managed, isolated Chrome session for live portal interaction, allowing Nova Act to automate tasks such as opening claims and triggering image analysis. This workflow separates repetitive screen work from domain reasoning, ensuring human expertise remains central to decision-making. The result is a more efficient intake process that preserves auditability and supports better routing and pattern analysis across the claims lifecycle. Source: awsml
During volume spikes from catastrophic events or seasonal surges, delays in FNOL processing can create backlogs that slow claim resolution. The current process requires claims professionals to manually validate evidence, often spending excessive time on repetitive tasks like locating artifacts and interpreting unstructured data. By automating these tasks, the new system allows adjusters to focus on higher-value decisions, improving both speed and consistency in claim handling. Source: awsml
