Building Secure AI Agents with Amazon Bedrock AgentCore and AWS Lambda MicroVMs

Explore how to build AI agents that can reason, interact with tools, and safely execute code. In this hands-on session, we’ll combine Amazon Bedrock AgentCore for agent orchestration, security, memory, and observability with AWS Lambda MicroVMs for isolated compute and sandboxed workload execution.

We’ll build a practical agent that can receive a task, determine when code execution is required, run the workload in an isolated environment, and use the results to complete the task.

You’ll learn:
* How Bedrock AgentCore enables production-ready AI agents
* How Lambda MicroVMs provide isolated compute for agent workloads
* How to safely allow agents to execute code
* Security and least-privilege considerations
* Observability and troubleshooting of agent workflows
* Where this architecture fits in real enterprise applications

(Please bring your laptops).