Building Custom Production-Ready AI Agents on AWS (Part 2)

This session covers the technical requirements for deploying robust AI agents on AWS beyond the proof-of-concept stage. It explores fundamental architecture using the Strands framework, including agent loops, state and session management, prompt engineering, and tool integration. The presentation focuses on leveraging AWS Bedrock AgentCore to implement production-grade features such as runtime environments, memory, identity, and observability. A real-world demonstration illustrates the integration of these components to build a functional and maintainable AI agent.