EvolveMem: Self-Evolving Memory Architecture for LLM Agents

Researchers have introduced EvolveMem, a self-evolving memory architecture for LLM agents that adapts both stored knowledge and retrieval mechanisms. Unlike traditional systems where retrieval infrastructure remains fixed, EvolveMem uses an LLM-powered diagnosis module to analyze failure logs, identify root causes, and propose targeted configuration adjustments. A guarded meta-analyzer applies changes with automatic revert-on-regression and explore-on-stagnation safeguards, enabling closed-loop self-evolution through an AutoResearch process. This approach aims to improve long-term memory across multiple sessions.