What is agentic memory (a-mem)?

Agentic Memory (A-MEM)

Agentic memory refers to an LLM agent’s dynamic long-term memory system that can grow and reorganize itself, as demonstrated by A-MEM. Instead of flat blob memory, A-MEM uses a structured knowledge network of “memory notes" (inspired by Zettelkasten) where each new memory is stored with descriptors and linked bidirectionally to related past memories. These notes -- containing context, keywords, and tags -- form an evolving graph that the agent can traverse to recall relevant knowledge later. Adding a new memory can trigger updates to older notes (“memory evolution”), letting the system refine and re-index what it knows over time. This self-organizing memory may yield more contextually aware reasoning and was shown to boost task performance versus baseline vector-store retrieval methods. (paper).