agent-memory

To add long-term memory to a LangGraph agent, integrate memory tools and configure a Lakebase store, then pass the user_id and store via the RunnableConfig to persist user context across sessions.

Long-term memory in conversational agents is used to persist user context and preferences across sessions, enabling personalized responses and coherent interactions without losing previous checkpointed history.

Yes, LangGraph long-term memory requires a Lakebase store. You must configure this store alongside a memory_tools factory and supply it through the config to enable per-user memory retrieval.

You save and recall user preferences across LangGraph sessions by using memory tools to checkpoint data. A config supplies the specific user_id and store, allowing the agent to retrieve personalized history.

To enable per-user memory identities in LangGraph, you need a memory_tools factory, a Lakebase store, and a config that supplies user_id and store to all memory tools with proper validation and error handling.