How MRH Trowe enabled secure self-service AI agents in financial services
aws ·
The article describes how MRH Trowe, a German commercial and industrial insurance broker, provided approximately 400 employees with secure, self-service AI agents during its first month of production. The solution combines Strands Agents, Amazon Bedrock AgentCore, and LibreChat to support governed access, German data residency, compliance, and cost transparency. Its first production agent retrieves Microsoft Teams meeting details and transcripts, then produces structured minutes from a single German-language request. Microsoft Entra ID ensures that each request runs under the signed-in employee’s identity and can access only that employee’s authorized calendar and transcript data, while processing remains in the AWS Europe (Frankfurt) Region. The architecture uses private connectivity, containerized LibreChat services, multiple AWS data stores, and AgentCore-hosted agents exposed through API Gateway and Lambda. Infrastructure and token costs were approximately $14 per seat per month, with a projected path to reduce infrastructure costs by about 40 percent through right-sizing and scheduled scaling. MRH Trowe is also expanding the framework to use cases such as analyzing CRM and public data, and aims to have subject matter experts create and maintain 10–15 agents by the end of 2026.
aws