Tools and Agents

A language model on its own can only produce text. Give it tools, such as a search function, a database query or a code runner, and it can look things up and change things in the world. The model never runs anything itself: it proposes a call, your code checks and executes it, and the result goes back into the conversation.

You will start with that single exchange, function calling, then put it in a loop to get an agent that reasons, acts and observes until a task is done. You will compare fixed workflows with autonomous agents and learn the common patterns, connect tools through the Model Context Protocol, give agents short-term and long-term memory, and see when several agents beat one. The module ends with coding and computer-use agents and the practical work of making agents reliable: tool design, retries, checkpoints, guardrails and evaluating whole trajectories.