Giving a model tools isn't enough: an agent also needs to decide which tool to use and when, and to notice when something went wrong. This module covers the patterns that structure that decision-making. The best known is ReAct (reasoning + acting), where the agent alternates between a thought, an action, and an observation until it can answer.
The first lesson explains ReAct clearly and visually. The second zooms out to the wider set of agentic design patterns — reflection, tool use, planning, and multi-agent collaboration — so you can see where reasoning fits. The last lesson is hands-on: building a ReAct-style agent in Python so the loop stops being abstract.
As you watch, keep one question in mind: what does the agent do when a tool returns something unexpected? Good reasoning and planning is mostly about handling that moment well.
A clear walkthrough of the ReAct framework, which combines an LLM's reasoning with the ability to take actions, and shows how the think-act-observe loop plays out step by step.
Video 5minTina Huang's overview of the main agent design patterns — reflection, tool use, planning, and multi-agent systems — and how they work together.
Video 1hA hands-on build of a ReAct (reasoning + acting) agent using plain Python, so you can see how the loop, the tools, and the model's reasoning fit together in code.
Video 29min