Are We Ready For An Agent-Native Memory System?

Are We Ready For An Agent-Native Memory System?

• ID: 1343ef0f
• 原文链接: https://arxiv.org/abs/2606.24775
• PDF: https://arxiv.org/pdf/2606.24775v1
• 作者: Wei Zhou, Xuanhe Zhou, Shaokun Han, Hongming Xu, Guoliang Li, Zhiyu Li, Feiyu Xiong, Fan Wu
• 日期: 2026-06-23
• 更新: 2026-06-23
• 分类: agents
• 来源类型: paper
• 标签: agent-memory, data-management, benchmark, evaluation, system-design
• 质量评分: 4/5
• 抓取时间: 2026-07-30T04:18:47Z

中文导读

从数据管理视角系统评测 agent 记忆，提出四模块分解框架（表示/存储抽取检索/路由维护）跨 5 个基准 11 个数据集评测 12 套记忆系统 + 2 基线核心发现：没有单一架构通吃，效果取决于记忆结构与负载瓶颈的对齐程度；局部维护比全局重组更有成本效率代码开源在 github.com/OpenDataBox/MemoryData

为什么值得关注

12 套 agent 记忆系统实测：没有通用最优架构，记忆结构与负载瓶颈的对齐才是关键

• arXiv comment: Paper list available at: https://github.com/OpenDataBox/awesome-agent-memory. Source code available at: https://github.com/OpenDataBox/MemoryData

关键信息

• 论文标题：Are We Ready For An Agent-Native Memory System?
• 作者：Wei Zhou, Xuanhe Zhou, Shaokun Han, Hongming Xu, Guoliang Li, Zhiyu Li, Feiyu Xiong, Fan Wu
• arXiv：https://arxiv.org/abs/2606.24775
• 发布时间：2026-06-23
• arXiv 分类：cs.CL, cs.DB, cs.IR
• 关联标签：agent-memory, data-management, benchmark, evaluation, system-design

English Abstract

Memory for large language model (LLM) agents has rapidly evolved from simple retrieval-augmented mechanisms into a data management system that supports persistent information storage, retrieval, update, consolidation, and dynamic lifecycle governance throughout agent execution. Despite this evolution, existing evaluations still benchmark agent memory mainly through end-to-end task success metrics (e.g., F1, BLEU), while treating the underlying system as a monolithic black box. As a result, critical system-level concerns, including operational costs, architectural trade-offs across memory modules, and robustness under dynamic knowledge updates, remain insufficiently explored. In this paper, we present a systematic experimental study of agent memory from a data management perspective. We propose an analytical framework that decomposes agent memory into four core modules: memory representation and storage, extraction, retrieval and routing, and maintenance. Under this framework, we evaluate 12 representative memory systems and two reference baselines across five benchmark workloads spanning 11 datasets. Our extensive end-to-end evaluation shows that no single architecture dominates across all scenarios; instead, effectiveness depends heavily on how well the memory structure aligns with the workload bottleneck. Furthermore, through fine-grained ablation studies, we quantify their individual effects on representation fidelity, retrieval precision, update correctness, and long-horizon stability. Finally, we reveal cost-performance trade-offs under realistic workloads, showing localized maintenance is more cost-efficient than global reorganization. Based on these findings, we identify promising directions towards building truly agent-native memory systems. The code is publicly available at https://github.com/OpenDataBox/MemoryData.

English Summary

A systematic experimental study of agent memory from a data management perspective. Proposes an analytical framework decomposing agent memory into four core modules: representation/storage, extraction, retrieval/routing, and maintenance. Evaluates 12 representative memory systems and 2 baselines across 5 benchmark workloads spanning 11 datasets. Key findings: no single architecture dominates across all scenarios; effectiveness depends on how well memory structure aligns with the workload bottleneck. Fine-grained ablations quantify effects on representation fidelity, retrieval precision, update correctness, and long-horizon stability. Reveals cost-performance trade-offs showing localized maintenance is more cost-efficient than global reorganization. Code at github.com/OpenDataBox/MemoryData.

Obsidian Notes

• 内容由 opencli arxiv paper 拉取 arXiv 元数据与摘要生成。
• 中文导读与价值判断均锚定在条目已有摘要、论文摘要、作者、日期与分类信息上；未补充论文摘要之外的实验细节。