Get Hands-On with AI Agents
Deep-dive hands-on workshops where you write real code and ship real agents — from SDK-level skill building and multi-agent orchestration to production deployment on Kubernetes and no-code agentic automation.
Agent Skill with Strands Agent SDK and Amazon Bedrock AgentCore
Hands-on Workshop 1 [WRK-001] | Level 200 – Intermediate | Laptop required
Stop reading about AI agents. Start building one. In this hands-on workshop, you’ll design, code, and deploy a production-ready AI agent skill using the Strands Agent SDK and Amazon Bedrock AgentCore. Define tool integrations, test skill execution in a managed runtime, and walk out with a working agent skill you can apply to your own business workflows on Day 1.
What you’ll build
• A fully functional agent skill running on Amazon Bedrock AgentCore
• Tool integrations that connect your agent to real-world systems
• A composable skill architecture you can extend into multi-agent systems
• Confidence to ship agent-powered automation at your organisation
Build AI agents that actually work: From concepts to production on Amazon Bedrock AgentCore
Hands-on Workshop [WRK-002] | Level 300 – Advanced | Laptop required
Chatbots answer questions. Agents close deals, validate strategies, and execute workflows. In this advanced workshop, you’ll go from zero to a deployed multi-agent application on Amazon Bedrock AgentCore — learning how agents reason, invoke tools via MCP, and collaborate to solve complex problems. Build incrementally: start with a single-agent loop, layer in tool and MCP server integrations, then orchestrate multiple agents that work together like a quant team at a hedge fund.
What you’ll build
- A single-agent reasoning loop with tool invocation
- MCP server integrations for secure, extensible tool access
- A multi-agent orchestration that validates trading strategies end-to-end
- A production deployment pattern you can replicate for your own use cases
Building production-ready AI agents on Amazon EKS
Hands-on Workshop [WRK-003] | Level 300 – Advanced | Laptop required
Most agent tutorials stop at the demo. This workshop starts where they end — on Kubernetes. Deploy model inference on accelerated hardware, wire vector-backed memory so your agent retains context across conversations, add secure tool access, and instrument everything for real observability. Two parallel tracks let you compare a fully self-managed stack versus a managed-service architecture — same agent, different infrastructure trade-offs. Leave with a clear decision framework for where to draw the build-vs-buy line.
What you’ll build
- AI agent inference running on accelerated hardware in Amazon EKS
- Persistent vector memory for context-aware conversations
- End-to-end observability and secure tool-access patterns
- A side-by-side comparison of self-managed vs. managed architectures
Build agentic automations for business processes with Amazon Quick Automate
Hands-on Workshop [WRK-004] | Level 200 – Intermediate | Laptop required
What if you could ship an agentic automation before your next standup? In this workshop, you’ll use Amazon Quick Automate’s chat-based authoring and visual studio to design, build, and deploy an end-to-end multi-agent workflow that coordinates AI agents across enterprise tools. Configure roles, wire tool integrations, debug execution paths with built-in observability, and deploy with enterprise-grade access controls — all without writing a single line of infrastructure code.
What you’ll build
- A multi-agent workflow orchestrating tasks across enterprise tools
- Agent roles and tool integrations configured via visual studio
- Observability and debugging for every execution path
- A production deployment with guardrails and access controls baked in
