How Amazon uses AI agents to support compliance screening of billions of transactions per day
Machine Learning Blog
This article describes how Amazon built an AI-powered compliance screening system using intelligent agents to process billions of transactions daily while maintaining regulatory compliance.
- Amazon screens ~2 billion transactions daily across 160+ businesses using AI agents
- Three-tier system: screening engine, ML filtering, and AI-powered investigation agents
- Specialized agents handle name matching, address matching, entity type inference, and verification
- Built on Strands framework, Amazon Bedrock LLMs, and Bedrock AgentCore Runtime
- Achieves 96% accuracy with 96% precision and 100% recall, automating 60% of cases
- Orchestration agent coordinates workflow; human review only for edge cases
- Key lessons: clear SOPs, iterative development, balanced autonomy, tool design quality, observability
Amazon's agentic AI system transforms compliance investigations at scale, automating routine cases while maintaining audit trails and regulatory adherence through deterministic agent orchestration.
The AWS News Feed is currently looking for gold sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.
Related articles
2025
2026
2026
2025
The AWS News Feed is currently looking for silver sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.