LangChain vs. LangGraph

Video: LangChain vs LangGraph: A Tale of Two Frameworks

LangChain = What happens inside each node (LLM prompt → output → tool call)
LangGraph = How nodes pass messages & state (graph scheduler + checkpoints)

Both coexist: every LangGraph node is still a LangChain Runnable, so moving from linear chains to a stateful graph is mostly about adding the orchestration layer, not rewriting the business logic.

Typical Workflows

LangChain Workflow

graph TB
    A[Input] --> B[Prompt Template]
    B --> C[LLM]
    C --> D[Output Parser]
    D --> E[Final Output]
    
    style A fill:#e1f5fe
    style E fill:#c8e6c9

LangGraph Workflow

graph TB
    A[Input State] --> B[Agent Node]
    B --> C{Decision}
    C -->|Tool Needed| D[Tool Node]
    C -->|Continue| E[LLM Node]
    D --> F[State Update]
    E --> F
    F --> G{More Steps?}
    G -->|Yes| B
    G -->|No| H[Final State]
    
    style A fill:#e1f5fe
    style H fill:#c8e6c9
    style C fill:#fff3e0
    style G fill:#fff3e0