So far a single request has produced a single answer. Many useful products need more: several steps, calls to tools, and decisions about what to do next. This module covers the two ways to build that, workflows and agents, and how to choose between them.
In a workflow, your code decides the order of the steps and the model fills in each one, which makes the result predictable. In an agent, the model decides which tools to use and when to stop, which is more flexible but harder to control. The lessons start with the core ideas, move to practical design principles such as keeping things simple and only adding an agent when a workflow will not do, and end with building these patterns in code.
Watch the lessons in order, then continue on to AI Product Design & UX.
The basics of AI agents and agentic workflows with simple, practical examples, and how they go beyond a chat assistant that just answers questions.
A beginner-friendly look at the real difference between an agent and a workflow, and why the distinction matters when you design a system.
A conference talk on when to use a simple workflow instead of an autonomous agent, keeping the architecture simple, and designing from the agent’s point of view.
Implementing the workflow and agent patterns from Anthropic’s “Building Effective Agents” post in code with LangGraph.