Build Your Own AI Agent

AI agents are reshaping how we build software — but most developers haven't built one themselves. In this hands-on half-day workshop, you'll go from zero to a working, tool-using AI agent that can reason about problems, take autonomous actions, and interact with files, code, and external APIs.

We'll work in three stages, each producing a complete, working agent you can take home:

Stage 1 — Setup and your first agent. You'll install Claude Code, connect your Anthropic API key, and scaffold a minimal agent from the CLI. You'll see how the agent loop works from the inside — how Claude reasons about a task, picks tools, acts, and iterates. By the end of this stage, you'll have a working agent that can read and reason about a codebase.

Stage 2 — Custom tools. You'll learn to extend your agent with custom tools — giving it new capabilities like calling external APIs, manipulating files, and searching the web. This is where it clicks: you'll watch your agent autonomously decide which tools to use, chain actions together, and recover from failures.

Stage 3 — Enhance a real project. You'll pick one of three prepared GitHub repositories — a web app with bugs to find, an API with missing documentation, or a data pipeline with failing tests — and use your agent to actually improve it. Read the code, diagnose issues, generate fixes, write tests.

You'll work with Claude Code the way you would on a real project, and walk away with a workflow you can use on Monday.