MemArchitect: A Policy Driven Memory Governance Layer
This addresses a critical governance gap for developers and users of autonomous LLM systems, though it is incremental as it builds on existing RAG frameworks.
The paper tackled the problem of unmanaged memory in persistent LLM agents, which leads to contradictions, privacy issues, and outdated information, by introducing MemArchitect, a governance layer that enforces policies like memory decay and conflict resolution, resulting in consistent performance improvements over unmanaged memory in agentic settings.
Persistent Large Language Model (LLM) agents expose a critical governance gap in memory management. Standard Retrieval-Augmented Generation (RAG) frameworks treat memory as passive storage, lacking mechanisms to resolve contradictions, enforce privacy, or prevent outdated information ("zombie memories") from contaminating the context window. We introduce MemArchitect, a governance layer that decouples memory lifecycle management from model weights. MemArchitect enforces explicit, rule-based policies, including memory decay, conflict resolution, and privacy controls. We demonstrate that governed memory consistently outperforms unmanaged memory in agentic settings, highlighting the necessity of structured memory governance for reliable and safe autonomous systems.