AWS has introduced a market surveillance agent that leverages LangGraph and Strands on AgentCore to manage complex workflows in financial services. The solution enables organizations to coordinate multiple specialized agents for real-time analysis of trading patterns and compliance monitoring. This system is designed to handle dynamic decision-making and robust error recovery, essential for production-ready agentic AI systems. The release of Amazon Bedrock AgentCore last year has simplified the deployment of such solutions, offering a foundation for scalable and observable AI applications. The complete implementation is available on GitHub for developers to explore and implement. Source: awsml
The system integrates LangGraph for workflow orchestration and Strands for intelligent agent reasoning. LangGraph provides a framework for managing state and directed graphs for multi-agent coordination, offering fine-grained control over execution and state sharing. It supports human-in-the-loop interactions and checkpoint-based recovery from failures, ensuring reliability in production environments. Strands Agent, on the other hand, serves as a reasoning engine within workflow nodes, offering model-agnostic capabilities that integrate with various large language models (LLMs) while maintaining flexible tool integration and observability. The combination allows for the creation of sophisticated multi-step analysis workflows that can adapt based on intermediate results. Source: awsml
The financial services industry faces unique challenges in market surveillance, requiring systems that can analyze trading patterns, detect suspicious activities, and generate reports while maintaining compliance standards. The solution addresses these challenges by separating data discovery from retrieval to prevent hallucinations and injection attacks. It uses tools like get_report_list and run_report to validate filters against report schemas and execute parameterized SQL queries without exposing raw SQL to the LLM. This approach ensures data integrity and security while enabling efficient and accurate market analysis. Source: awsml
