Welcome to Agent Foundations, the first module of the course. Before you write a single tool or loop, it’s worth getting crisp about what an “agent” really is — because the word is used loosely, and most of the confusion around agents comes from skipping this step. An agent isn’t a bigger prompt or a smarter model. It’s a pattern: a language model placed inside a loop where it chooses its own next action, uses tools, sees the results, and keeps going until the goal is met.

In this module you’ll build that mental model precisely. You’ll learn the agent loop that everything else rests on, see clearly how an agent differs from a workflow (and when each is the right choice), break down the anatomy of an agent into the handful of parts every one shares, and make your first call to Claude with the Anthropic SDK. The module closes with a guided design exercise: planning Atlas, the trip-planning agent you’ll build across the rest of the course.

The Claude API code uses the low-cost claude-haiku-4-5 model, and the orchestration you’ll write later is real, runnable Python. Start with Lesson 1, where you’ll learn exactly what an agent is.