Building an AI agent that works in a demo is straightforward. Deploying one that performs reliably in production, with real users and unpredictable inputs, requires a different level of engineering. This session walks through the complete journey from a simple prototype to a production-ready AI agent using the open-source Strands Agents SDK and Amazon Bedrock AgentCore Runtime.

The session demonstrates how to build an agent with minimal code, using a system prompt, a model, and tools, and then focuses on the critical steps often overlooked in tutorials: packaging, deployment, scaling, security, and monitoring in a managed cloud environment. It also examines a model-driven approach to agent design, where modern LLM capabilities reduce the need for heavy orchestration frameworks. Through a live walkthrough, you will see how an agent can be taken from a local setup to a fully deployed, scalable endpoint on AWS, with support for concurrency, long-running tasks, and operational resilience.

What You Will Learn

• How to build and deploy an AI agent using Strands Agents SDK and Amazon Bedrock AgentCore Runtime

• How to handle scaling, concurrency, and long-running tasks in production environments

• Key considerations for packaging, securing, and monitoring AI agents in the cloud

Who Should Attend

• Backend developers

• Cloud and platform engineers

• AI and machine learning practitioners

• Software architects evaluating AI agent deployment strategies