![[2602.19320] Anatomy of Agentic Memory: Taxonomy and Empirical Analysis of Evaluation and System Limitations](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.19320] Anatomy of Agentic Memory: Taxonomy and Empirical Analysis of Evaluation and System Limitations
Summary
This article presents a comprehensive analysis of agentic memory systems in large language models, highlighting their architectural frameworks and empirical limitations.
Why It Matters
Understanding agentic memory is crucial for enhancing the performance of large language models, which are increasingly used in AI applications. This analysis identifies key challenges and suggests improvements, making it relevant for researchers and developers in AI and machine learning.
Key Takeaways
- Agentic memory systems support long-horizon reasoning and personalization in LLMs.
- Current evaluation metrics and benchmarks are often misaligned with actual performance.
- System limitations include benchmark saturation, metric validity, and backbone-dependent accuracy.
- A structured taxonomy of memory systems is proposed to clarify architectural differences.
- The paper outlines directions for improving evaluation methods and system design.
Computer Science > Computation and Language arXiv:2602.19320 (cs) [Submitted on 22 Feb 2026] Title:Anatomy of Agentic Memory: Taxonomy and Empirical Analysis of Evaluation and System Limitations Authors:Dongming Jiang, Yi Li, Songtao Wei, Jinxin Yang, Ayushi Kishore, Alysa Zhao, Dingyi Kang, Xu Hu, Feng Chen, Qiannan Li, Bingzhe Li View a PDF of the paper titled Anatomy of Agentic Memory: Taxonomy and Empirical Analysis of Evaluation and System Limitations, by Dongming Jiang and 10 other authors View PDF HTML (experimental) Abstract:Agentic memory systems enable large language model (LLM) agents to maintain state across long interactions, supporting long-horizon reasoning and personalization beyond fixed context windows. Despite rapid architectural development, the empirical foundations of these systems remain fragile: existing benchmarks are often underscaled, evaluation metrics are misaligned with semantic utility, performance varies significantly across backbone models, and system-level costs are frequently overlooked. This survey presents a structured analysis of agentic memory from both architectural and system perspectives. We first introduce a concise taxonomy of MAG systems based on four memory structures. Then, we analyze key pain points limiting current systems, including benchmark saturation effects, metric validity and judge sensitivity, backbone-dependent accuracy, and the latency and throughput overhead introduced by memory maintenance. By connecting the memo...
