💾
Memory Patterns
Approaches to giving agents persistent knowledge — from conversation history to long-term semantic memory.
4 patterns
Retrieval-Augmented Generation
Intermediate
Retrieve relevant information from an external knowledge base before generating a response, grounding the LLM in facts.
Long-Term Memory
Intermediate
Persist information across sessions so agents can remember user preferences, past interactions, and learned knowledge.
Episodic Memory
Advanced
Store and retrieve specific past experiences — complete interaction episodes — so the agent can learn from what worked and what didn't.
Working Memory
Basic
Maintain a scratchpad of relevant information within a single task execution, actively curating what stays in the agent's context window.