AI Agents in LangGraph
by DeepLearning.AI × LangChain
Build an agent from scratch, then rebuild it in LangGraph to see how a graph runtime adds control and persistence.
Overview
Taught with Harrison Chase (LangChain) and Andrew Ng, this course first has you implement a ReAct-style agent in plain Python so you understand the mechanics, then reconstructs it in LangGraph. You learn to model agents as graphs with explicit state, add persistence and checkpoints, stream tokens, and insert human approval steps. It's the fastest way to understand why a graph-based runtime beats an ad-hoc while-loop for real agents.
At a Glance
- Topic
- Agentic
- Level
- Intermediate
- Format
- Course
- Cost
- Free
- Duration
- ~1-2 hours
- Provider
- DeepLearning.AI × LangChain
- Hands-on
- Yes — code/exercises
- Certificate
- None
What You’ll Learn
- ✓Implementing the ReAct agent loop from scratch
- ✓Modeling agents as stateful graphs in LangGraph
- ✓Persistence, checkpoints, and streaming
- ✓Adding human-in-the-loop approval
Highlights
- •Co-taught by LangChain's founder
- •From-scratch first, framework second
- •Free while in the DeepLearning.AI catalog
Who It’s For
Best For
- ✓Developers building stateful, controllable agents
Prerequisites
- •Intermediate Python
- •Basic LLM API experience
FAQ
What is AI Agents in LangGraph?
A short, code-along course that teaches the agent loop from scratch and then shows how LangGraph adds cyclic control flow, state, and human-in-the-loop.
Is AI Agents in LangGraph free?
AI Agents in LangGraph is free to access.
What level is AI Agents in LangGraph for?
AI Agents in LangGraph is aimed at a intermediate audience. Recommended background: Intermediate Python, Basic LLM API experience.
How long does AI Agents in LangGraph take?
Expect roughly ~1-2 hours. Most learners work through it at their own pace.
What will I learn from AI Agents in LangGraph?
You'll learn: Implementing the ReAct agent loop from scratch; Modeling agents as stateful graphs in LangGraph; Persistence, checkpoints, and streaming; Adding human-in-the-loop approval.