How Amazon Finance streamlines regulatory inquiries by using generative AI on AWS

This article describes how Amazon Finance Technology built a scalable AI system using Amazon Bedrock to automate regulatory inquiry processing and response generation.

• Uses Amazon Bedrock Knowledge Bases with OpenSearch Serverless for document retrieval
• Implements multi-turn conversations with Claude Sonnet 4.5 via Converse Stream API
• Stores conversation history in DynamoDB for contextual awareness across sessions
• Employs query expansion strategy to handle acronyms and abbreviated user questions
• Parallel retrieval processing reduced latency from 10 seconds to under 2 seconds
• Integrates Amazon Bedrock Guardrails for PII and financial data protection
• Uses OpenTelemetry and self-hosted Langfuse for comprehensive observability
• Hierarchical chunking strategy optimizes document indexing for complex financial documents

The solution demonstrates a production-grade pattern for building secure, compliant, and scalable AI-powered knowledge workflows on AWS serverless infrastructure.