SuperLocalMemory V3 establishes information-geometric foundations for agent memory, enabling high-accuracy retrieval without cloud-based LLM dependency.
March 17, 2026
Original Paper
SuperLocalMemory V3: Information-Geometric Foundations for Zero-LLM Enterprise Agent Memory
arXiv · 2603.14588
The Takeaway
It replaces standard cosine similarity with a metric derived from Fisher information and uses sheaf-theoretic models to detect irreconcilable contradictions in memory. It provides a path to architecturally private, zero-LLM enterprise agents that comply with strict data sovereignty laws.
From the abstract
Persistent memory is a central capability for AI agents, yet the mathematical foundations of memory retrieval, lifecycle management, and consistency remain unexplored. Current systems employ cosine similarity for retrieval, heuristic decay for salience, and provide no formal contradiction detection.We establish information-geometric foundations through three contributions. First, a retrieval metric derived from the Fisher information structure of diagonal Gaussian families, satisfying Riemannian