Build real AI agents that plan, use tools, and follow workflows with ReAct, ReWOO, and LangGraph.
AI Agents in Practice is a practical, beginner-friendly course that shows you how to design and build working agentic systems using today’s most relevant tools and frameworks, including ReAct, ReWOO, LangGraph, and LangSmith. It’s the natural next step for anyone who understands the basics of large language models and simple chatbots and now wants to build agents that can plan, use tools, and follow multi-step workflows.
What you’ll learn
- Master the core principles of agentic design..
- Gain a solid understanding of two leading agent frameworks: ReAct and ReWOO..
- Create and integrate tools that let agents work with real data and perform meaningful tasks..
- Develop confidence in crafting prompts tailored for agents—including reasoning, planning, and tool use..
- Monitor and debug agents with LangSmith, so you always know what’s happening under the hood..
Course Content
- Introduction to the Course –> 1 lecture • 4min.
- Agentic Systems in Practice –> 8 lectures • 30min.
- Project 1 – Job-Helper agent (ReAct) –> 13 lectures • 29min.
- Project 2 – ReWOO Job-Helper agent –> 7 lectures • 21min.
- Project 3 – Business-Idea Evaluator –> 8 lectures • 20min.
Requirements
AI Agents in Practice is a practical, beginner-friendly course that shows you how to design and build working agentic systems using today’s most relevant tools and frameworks, including ReAct, ReWOO, LangGraph, and LangSmith. It’s the natural next step for anyone who understands the basics of large language models and simple chatbots and now wants to build agents that can plan, use tools, and follow multi-step workflows.
Along the way, we’ll tackle the questions most people have when they first encounter AI agents, such as:
- What drives an AI system browsing the web, reading files, or calling APIs to decide what to do next?
- In what way does it break a task into steps?
- How does it determine which tool to use?
- When does it know to ask a human for help?
If you want clear, practical answers to these questions without getting lost in theory, this course is for you.
We begin with a concise introductory section that provides a solid understanding of what an AI agent is, how it differs from a standard LLM application, and how agents are used in real projects.
- Grasp the core building blocks of an agent.
- See how agentic systems fit into real-world AI applications.
- Apply best practices for creating prompts and prompt frameworks.
- Understand how system and user messages shape agent behavior.
- Explore prompt patterns that guide an agent’s reasoning.
- Look behind the scenes of a real helper chatbot to connect each concept to a concrete example.
In Project 1, you’ll build a Job-Helper agent using the ReAct pattern, turning theory into a working system step by step.
- Explore the structure of a LangGraph project.
- Create tools like a file reader and a web-search helper.
- Add memory so the agent can use information from earlier steps.
- Build and run the graph that ties everything together.
- Trace the agent’s behavior in LangSmith.
In Project 2, you’ll create a new version of the Job-Helper agent using ReWOO, giving you a hands-on comparison of two agentic architectures.
- Shift from the ReAct pattern to ReWOO.
- Define the planner, executor, and solver nodes in LangGraph.
- Compare both approaches in LangSmith, examining latency, cost, and behavior.
In Project 3, you’ll bring everything together in a new project called the Business Idea Evaluator, a richer workflow that combines multiple techniques.
- Build advisor “personas” that evaluate ideas from different perspectives.
- Combine two powerful methods: human-in-the-loop steps for adding context, and parallelization to speed up evaluation.
- Use a final collection node to merge all outputs into a single, clear assessment.
By the end of the course, you’ll understand:
- How modern agents think and operate.
- The differences between ReAct and ReWOO differ, and when to use each.
- Techniques for designing prompts that support reasoning, planning, and tool use.
- How to structure an agent as a LangGraph with nodes, edges, state, and memory.
- Ways to integrate custom tools and external APIs into your graph.
- Methods for adding human-in-the-loop stages and parallel branches to your workflows.
- How to monitor and debug your agents with LangSmith instead of working blindly
We break down complex concepts and code into small, digestible steps that make it easy to follow along and start building. Whether you want to expand your portfolio, level up your AI skills, or simply understand how real agents work under the hood, this course is designed to help you make that leap with confidence.