Agent Memory: Structure, Time, and Evaluation
Research map for persistent agent memory: semantic structure, temporal graphs, episodic memory, memory operating systems, and long-term evaluation.
Paper • 2603.08965 • Published • 1Note Mnemoverse research on continuous abstraction control and emergent scale boundaries in knowledge graphs.
LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory
Paper • 2410.10813 • Published • 16Note An evaluation frame for extraction, multi-session reasoning, temporal reasoning, updates, and abstention.
Zep: A Temporal Knowledge Graph Architecture for Agent Memory
Paper • 2501.13956 • Published • 16Note A production-oriented temporal knowledge graph approach and an important comparison point for dynamic memory.
AriGraph: Learning Knowledge Graph World Models with Episodic Memory for LLM Agents
Paper • 2407.04363 • Published • 34Note Connects semantic and episodic memory in a graph world model for exploration and planning.
MemOS: A Memory OS for AI System
Paper • 2507.03724 • Published • 170Note Treats memory as a managed system resource spanning plaintext, activation, and parameter-level representations.
Graph-based Agent Memory: Taxonomy, Techniques, and Applications
Paper • 2602.05665 • PublishedNote A recent taxonomy mapping graph-based agent-memory design choices across structures, operations, and evaluation.
Harness the Memory: A Holistic Evaluation of Memory Substrates in Memory Agents
Paper • 2608.15008 • Published • 15Note Controlled evaluation of diverse memory substrates across task and scale regimes; shows that no single substrate dominates and motivates adaptive routing.