All-Mem: Agentic Lifelong Memory via Dynamic Topology Evolution

TL;DR

All-Mem introduces a dynamic, topology-structured lifelong memory system for interactive agents, improving long-term retrieval accuracy and efficiency through explicit consolidation and bounded search strategies.

Contribution

It proposes a novel topology-evolving memory framework with explicit consolidation and hop-bounded retrieval, addressing degradation issues in existing systems.

Findings

Enhanced retrieval accuracy on LOCOMO and LONGMEMEVAL datasets

Maintains traceability of evidence through immutable links

Achieves bounded search cost with a structured memory topology

Abstract

Lifelong interactive agents are expected to assist users over months or years, which requires continually writing long term memories while retrieving the right evidence for each new query under fixed context and latency budgets. Existing memory systems often degrade as histories grow, yielding redundant, outdated, or noisy retrieved contexts. We present All-Mem, an online/offline lifelong memory framework that maintains a topology structured memory bank via explicit, non destructive consolidation, avoiding the irreversible information loss typical of summarization based compression. In online operation, it anchors retrieval on a bounded visible surface to keep coarse search cost bounded. Periodically offline, an LLM diagnoser proposes confidence scored topology edits executed with gating using three operators: SPLIT, MERGE, and UPDATE, while preserving immutable evidence for traceability. At query time, typed links enable hop bounded, budgeted expansion from active anchors to archived evidence when needed. Experiments on LOCOMO and LONGMEMEVAL show improved retrieval and QA over representative baselines.