AI Agents Tutorial

This course is for developers who have called a language model and now want to build something that acts on its own: look things up, call APIs, and decide what to do next. You should be comfortable reading Python and know roughly what a language model does. If not, work through the Large Language Models and Prompt Engineering courses first.

The approach is vendor-neutral. Every example uses a plain Python function as a stand-in for the model, so the programs run offline and print the same output every time. That lets you study the parts that are really yours to design: the loop, the tool layer, memory and state, and the safety rails. Once those are clear, swapping in a real model from any provider is a small change.

The four lessons go from the core loop, to how tools are described and validated, to memory and planning, and finally to reliability: budgets, human approval, evaluation and the typical ways agent systems fail.