Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MemArchitect: A Policy Driven Memory Governance Layer

Lingavasan Suresh Kumar, Yang Ba, Rong Pan

Abstract

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frameworks treat memory as passive storage, lacking mechanisms to resolve contradictions, enforce privacy, or prevent outdated information ("zombie memories") from contaminating the context window.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.

MemArchitect: A Policy Driven Memory Governance Layer

Abstract

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frameworks treat memory as passive storage, lacking mechanisms to resolve contradictions, enforce privacy, or prevent outdated information ("zombie memories") from contaminating the context window.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.
Paper Structure (28 sections, 9 equations, 1 figure, 1 table)

This paper contains 28 sections, 9 equations, 1 figure, 1 table.

Table of Contents

  1. Introduction
  2. The Governance Engine: Policy Layer
  3. The Policy Suite
  4. The Governance Cycle & Dependencies
  5. Experiment & Analysis
  6. Experiment Setup and Baseline
  7. Quantitative Results
  8. Conclusion and Future Work
  9. GOVERNANCE LAYER: THE COMPLETE POLICY LAYER
  10. Domain A: Lifecycle & Hygiene (The "Forgetting" Engine)
  11. Policy 1: FSRS Retrievability (Active)
  12. Policy 2: Stability Update (Active)
  13. Policy 3: Entropy-Triggered Consolidation (Active)
  14. Domain B: Consistency & Truth (The "Utility" Engine)
  15. Policy 4: Kalman Utility Filter (Active)
  16. ...and 13 more sections

Figures (1)

  • Figure 1: The MemArchitect Governance Workflow. The system operates as a continuous closed loop across three paths: (1) Read Path (Top): Incoming queries are classified by intent to customize the search. Memories are retrieved, filtered for safety, and ranked by value to fit the limited context budget. (2) Reflect Path (Right): A feedback loop detects which memories were actually used to generate the answer, strengthening useful information while penalizing distractions. (3) Background Path (Bottom): During idle cycles, a maintenance process automatically cleans the memory store by deleting forgotten noise and compressing fading details into permanent facts.