Create an intelligent insurance underwriter agent powered by Amazon Nova 2 Lite and Amazon Quick Suite

Insurance underwriting requires analyzing multiple data sources, evaluating risks, and making decisions that meet regulatory requirements. The underwriters face three core challenges:

• Siloed data scattered across Customer Relationship Management (CRM) systems, document repositories, and transactional databases
• Regulatory requirements for explainable, auditable decisions that traditional black box AI can’t satisfy
• The need for consistent, automated underwriting rules with proactive fraud detection across portfolios

In this post, we show how to overcome these challenges with an enterprise-grade underwriting solution using Amazon Nova 2 Lite that unifies your data sources and delivers audit-ready risk assessments.

Solution overview

The solution uses Model Context Protocol (MCP) tools for insurance fraud detection, risk assessment of an applicant, and underwriting decisions. When a tool is invoked based on a user query, the tool fetches data from required data sources—document repositories like Amazon Simple Storage Service (Amazon S3) and databases like Amazon DynamoDB—and calls the ...