Course Overview
This short course walks you through the mechanics of building an AI agent twice — first by hand, then again using LangGraph, LangChain's framework for creating more structured and controllable agent workflows. The idea is to show you exactly what a framework is doing for you under the hood before you start relying on it. Harrison Chase, who co-founded LangChain, teaches alongside Rotem Weiss, co-founder of Tavily, a search tool built specifically for feeding agents clean, structured results instead of raw links. Between them, you get both the architecture side and the data-retrieval side of agent-building.
The course doesn't stop at a basic agent loop. It pushes into practical production concerns like saving agent state so you can pause and resume conversations, adding human review checkpoints before an agent takes action, and streaming outputs as they're generated. The final project — an essay-writing agent — ties these pieces together into something resembling a real research assistant workflow.
What You Will Learn
- How to construct an agent from scratch in Python before introducing any framework
- The core building blocks of LangGraph and how they fit together
- How agentic search differs from standard search engines by returning structured, agent-ready answers
- Techniques for persisting agent state across sessions and switching between conversation threads
- How to insert human-in-the-loop checkpoints so a person can approve or redirect agent decisions
- How to build a multi-step essay-writing agent that mimics a researcher's workflow
Course Structure
The course runs through 9 lessons combining video with 6 embedded code examples, plus a graded quiz:
- Introduction (6m)
- Build an Agent from Scratch (12m)
- LangGraph Components (19m)
- Agentic Search Tools (5m)
- Persistence and Streaming (9m)
- Human in the Loop (14m)
- Essay Writer (18m)
- LangChain Resources (2m)
- Conclusion (4m)
- Graded Quiz (10m)
Who Is This Course For?
This is built for people who already write Python comfortably and want to move from basic LLM API calls into agent design. If you've never written a function or don't know what a loop is, you'll be lost fast — the course assumes you can keep up with code examples without hand-holding. On the other hand, if you've already shipped agents with LangGraph in production, the content here is likely too introductory to teach you much new.
Format & Time Commitment
Everything is self-paced video with inline coding exercises — there are no deadlines or live sessions. At under two hours total, it's realistic to finish in a single sitting or spread across two or three short sessions.
Pros and Cons
Pros
- Taught directly by the creators of LangChain and Tavily, so the explanations come from people who built the tools
- Covers production-relevant topics like persistence and human-in-the-loop, not just toy examples
- Short enough to complete without a major time investment
Cons
- At under two hours, depth is limited — this is an orientation, not mastery
- Requires existing Python comfort, so it's not accessible to true beginners
- The official certificate/accomplishment sits behind a PRO membership, so finishing the free version alone doesn't get you a credential
- Free access is explicitly tied to a platform beta period, meaning pricing could change later
FAQ
Is this course free?
Yes, access is currently free while DeepLearning.AI's platform is in beta, though that may not be permanent.
Do I need prior experience with LangChain?
No, but you do need intermediate Python skills — the course builds LangGraph concepts from the ground up.
Will I get a certificate?
Only if you have a PRO membership; the accomplishment/credential isn't included in the free tier.
What if I want something more in-depth afterward?
This course works well as a primer before exploring LangChain's full documentation or longer agent-focused courses, since it only covers roughly 1h42m of material.
If you want a fast, instructor-led primer on LangGraph straight from its creator, it's worth checking the official DeepLearning.AI course page before deciding.
