← Learning
LangGraph
A linear, one-concept-per-lesson path through LangGraph, the graph engine LangChain agents are built on, from your first graph to a multi-agent research assistant. 35 lessons · 3 tiers.
Prerequisites: Completion of LangChain & Agent Building, or already comfortable with .invoke(), messages, and @tool. This course does not re-teach those.
Lessons use Google's Gemini free tier (gemini-3.5-flash-lite), the same model as the LangChain course, for the same reason: no prepaid API credits needed to learn.
Built on LangGraph, the open-source framework this course teaches.
Beginner
Graph primitives, no memory or tools yet.
- 01First GraphGitHub
- 02State and ReducersGitHub
- 03Multiple Nodes, LinearGitHub
- 04Conditional EdgesGitHub
- 05Cycles and LoopsGitHub
- 06MessagesStateGitHub
- 07Tool NodeGitHub
- 08Streaming: Values and UpdatesGitHub
- 09Streaming TokensGitHub
- 10Invoke Config and RecursionGitHub
- 11Visualizing the GraphGitHub
- 12Checkpoint: Text-Processing Pipeline GraphGitHub
Intermediate
Memory, persistence, parallelism, control flow.
- 13Checkpointer MemoryGitHub
- 14Persistent CheckpointerGitHub
- 15Human in the Loop: interrupt()GitHub
- 16Time Travel: State HistoryGitHub
- 17Parallel Fan-Out, Fan-InGitHub
- 18Send API: Map-ReduceGitHub
- 19SubgraphsGitHub
- 20Command ObjectsGitHub
- 21Retry and Error HandlingGitHub
- 22Checkpoint: Persistent Approval WorkflowGitHub
Advanced
Real agents, multi-agent systems, production concerns.
- 23Building an Agent From ScratchGitHub
- 24Structured Output in a GraphGitHub
- 25Long-Term Memory StoreGitHub
- 26Multi-Agent HandoffsGitHub
- 27Multi-Agent Supervisor GraphGitHub
- 28Context Trimming in a NodeGitHub
- 29RAG NodeGitHub
- 30Dynamic BreakpointsGitHub
- 31Middleware-Style GuardrailsGitHub
- 32Tracing and ObservabilityGitHub
- 33Async and ConcurrencyGitHub
- 34Deploying a Persistent AgentGitHub
- 35Capstone: Multi-Agent Research Assistant GraphGitHub