A hands-on day where you'll build an AI agent with Claude Code and deploy it on Amazon Bedrock AgentCore.

To run an agent in production, you need more than the agent code. You also need hosting, memory, tool access, user auth, and monitoring. AgentCore provides each of these as a managed service.

We'll start with a walkthrough of the AgentCore features you can use to host your agents:

• Runtime - hosts your agent and isolates each user session. Works with any framework (e.g., Strands Agents, LangGraph) or model.
• Memory - stores short-term and long-term memory, so your agent remembers context across sessions
• Gateway - turns your existing APIs and Lambda functions into MCP tools your agent can call
• Identity - lets your agent sign in to other services (e.g., GitHub, Slack) on behalf of your users
• Code Interpreter & Browser - sandboxed tools for running code and using websites
• Observability - shows each step your agent takes in Amazon CloudWatch, so you can debug it

Then you'll use Claude Code to build an agent and deploy it to AgentCore. By the end of the day, you'll have your agent running on AWS.

We'll provide workshop accounts with Claude on Amazon Bedrock already set up. All Claude Code usage runs through these accounts, so you don't need your own AWS account or API key.

Who it's for: developers and technical founders building agents (or planning to). Some Python experience will help.

What to bring: a laptop.