LangGraph is a framework for building sophisticated, stateful AI agents. Its graph-based architecture supports flexible workflows, complex interactions, state management, and multi-turn conversations, while integrating with LangChain. section 9.2.3
Evaluate LangGraph as both a capability and an engineering trade-off:
- Use graph-based workflows when agents must preserve state or manage complex interactions across turns. section 9.2.3
- Compare its abstraction and overhead with alternatives before committing, since agent frameworks can reduce code flexibility and create lock-in. section 9.2.3
Figure 9.14 Agents need short-term memory for the current session and long-term memory for learning and improvement over time.
Use the memory distinction to relate LangGraph’s stateful agents to short-term session context and longer-term learning or improvement. section 9.2.3
