Cross-Session Persistence
How to build agents whose memory survives restarts - architecture, storage backends, session restoration, and privacy-aware memory pruning for production systems.
7 docs tagged with "agent-memory"
View all tagsEpisodic Memory with Vector Store
Implement agent episodic memory using vector databases: storing, retrieving, consolidating, and forgetting past experiences at scale.
Four Types of Agent Memory
Cognitive science meets AI engineering: working, episodic, semantic, and procedural memory implemented in production agent systems.
In-Context Working Memory
Managing the context window as working memory: token budgeting, sliding windows, summarization, and the lost-in-the-middle problem.
Memory Compression and Summarization
How to keep agents functional across days-long tasks by compressing memory intelligently - preserving what matters, discarding what does not.
Procedural Memory and Learned Skills
How agents store and reuse successful action sequences: skill formation, retrieval, composition, and refinement from execution feedback.
Semantic Memory and Knowledge Graphs
Structured world knowledge for agents: building and querying knowledge graphs with entity extraction, relationship traversal, and hybrid vector+graph retrieval.