A-Mem: Agentic Memory for LLM Agents
Open accessThe DOI matches its Crossref record; 33 of the 33 references it cites were found in a registry or an index; 11 of the 17 sources its claims cite were checked for support.
Abstract
This paper proposes a novel agentic memory system for LLM agents that can dynamically organize memories in an agentic way following Zettelkasten note-taking principles with dynamic indexing and linking.
Overview
CoChat Summary
TL;DR
A-MEM is an agentic memory system for LLM agents that, inspired by the Zettelkasten method, builds structured notes with LLM-generated keywords, tags, and contextual descriptions, then autonomously links and evolves them as new memories arrive. Across six foundation models on LoCoMo and DialSim, it outperforms LoCoMo, ReadAgent, MemoryBank, and MemGPT while using far fewer tokens.
Key findings
- On LoCoMo with GPT-4o-mini, A-MEM reaches 27.02 F1 / 20.09 BLEU-1 on Multi Hop and 45.85 / 36.67 on Temporal, versus MemGPT's 26.65 / 17.72 and 25.52 / 19.44; A-MEM's average ranking is 1.2 versus 2.4 for both LoCoMo and MemGPT.
- On DialSim, A-MEM scores F1 3.45, BLEU-1 3.37, ROUGE-L 3.54, ROUGE-2 3.60, METEOR 2.05, SBERT 19.51, versus LoCoMo (2.55, 3.13, 2.75, 0.90, 1.64, 15.76) and MemGPT (1.18, 1.07, 0.96, 0.42, 0.95, 8.54) — a 35% F1 improvement over LoCoMo and 192% over MemGPT.
- Ablation with GPT-4o-mini: removing both link generation and memory evolution drops Multi Hop F1 to 9.65; removing only memory evolution gives 21.35; full A-MEM gives 27.02.
- Cost/scaling: ~1,200 tokens per memory operation (85–93% reduction vs. ~16,900 for LoCoMo/MemGPT), under $0.0003 per operation, 5.4 s average with GPT-4o-mini and 1.1 s with local Llama 3.2 1B; retrieval time grows only from 0.31 µs to 3.70 µs from 1,000 to 1,000,000 memories.
Methods
- Design: each memory note is `{content, timestamp, keywords, tags, context, embedding, links}`; an LLM generates keywords/tags/context, a text encoder (all-minilm-l6-v2) produces embeddings, top-k nearest neighbors are retrieved by cosine similarity, and an LLM decides link generation and memory evolution (updating neighbors' context, keywords, tags).
- Data: LoCoMo (7,512 QA pairs, dialogues averaging 9K tokens over up to 35 sessions; single-hop, multi-hop, temporal, open-domain, adversarial) and DialSim (multi-party dialogues from Friends, The Big Bang Theory, The Office; 1,300 sessions, ~350,000 tokens, >1,000 questions per session).
- Evaluation: F1 and BLEU-1 as primary metrics, plus ROUGE-L, ROUGE-2, METEOR, and SBERT similarity; baselines LoCoMo, ReadAgent, MemoryBank, MemGPT; six foundation models (GPT-4o-mini, GPT-4o, Qwen2.5-1.5B/3B, Llama 3.2 1B/3B) plus DeepSeek-R1-32B, Claude 3.0 Haiku, and Claude 3.5 Haiku in the appendix; top-k retrieval with k=10 default and category-specific tuning.
Limitations
- Stated by the paper: memory organization quality may still depend on the underlying LLM's capabilities, and different LLMs may generate different contextual descriptions or connections; the implementation is text-only, with multimodal (image/audio) extension left to future work.
- Evident from the text: no error bars or significance tests are reported (the checklist answers "No" to statistical significance, citing API cost); the main-text results are reported for a single dataset family of long-term conversation QA, and the paper's own checklist notes broader societal impacts are not discussed.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
5 Valid · 0 Not valid · 3 Not checked · 55 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“These systems require agent developers to predefine memory storage structures, specify storage points within the workflow, and establish retrieval timing.”
What the sources say
- Partly supportedcites [25] [25] Charles Packer, Sarah Wooders, Kevin Lin, Vivian Fang, Shishir G Patil, Ion Stoica, and Joseph E Gonzalez. Memgpt: Towards llms as operating systems. arXiv preprint arXiv:2310.08560, 2023. sourceExperimental — accuracy not yet measured
Section 3.2.1 (p. 6) says the fixed-context baselines fetch documents independently of LLM inference, supporting the idea that retrieval is set by the workflow rather than chosen by the agent. Section 2.3 (p. 3) contrasts this with MemGPT’s self-directed memory edits and retrieval. The paper does not establish that existing memory systems generally require developers to predefine storage structures or specify storage points. (LLM, conf=0.78)
- Could not verifycites [39] [39] Wanjun Zhong, Lianghong Guo, Qiqi Gao, He Ye, and Yanlin Wang. Memorybank: Enhancing large language models with long-term memory. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 19724–19731, 2024. 12 Contents sourceExperimental — accuracy not yet measured
Sections 2.1 and 2.2 describe MemoryBank’s storage structure and retrieval mechanism, but do not say that existing systems require agent developers to predefine structures, choose workflow storage points, or set retrieval timing. Searches for those developer requirements found no passage addressing them. (LLM, conf=0.73)
- Not matchedcites [28] [28] Aymeric Roucher, Albert Villanova del Moral, Thomas Wolf, Leandro von Werra, and Erik Kaunismäki. 'smolagents': a smol library to build great agentic systems. https://github. com/huggingface/smolagents, 2025.
Its entry has no title and year that we can search by.
- Could not verifycites [21] [21] Zhiwei Liu, Weiran Yao, Jianguo Zhang, Liangwei Yang, Zuxin Liu, Juntao Tan, Prafulla K Choubey, Tian Lan, Jason Wu, Huan Wang, et al. Agentlite: A lightweight library for building and advancing task-oriented llm agent system. arXiv preprint arXiv:2402.15538, 2024. sourceExperimental — accuracy not yet measured
Section 2.1 describes when AgentLite retrieves an action-observation history and stores an observation; the Figure 1 description in Section 6 also describes its memory workflow. Neither passage says that existing memory systems require developers to predefine storage structures, storage points, or retrieval timing. Searches for those requirements found no passage addressing them. (LLM, conf=0.72)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Existing memory systems [25, 39, 28, 21] for LLM agents provide basic memory storage functionality.”
What the sources say
- Supportedcites [25] [25] Charles Packer, Sarah Wooders, Kevin Lin, Vivian Fang, Shishir G Patil, Ion Stoica, and Joseph E Gonzalez. Memgpt: Towards llms as operating systems. arXiv preprint arXiv:2310.08560, 2023. source
MemGPT is an LLM-agent memory system that stores information outside the context window. Section 2 (p. 2) describes its external context and memory-management functions, and the Figure 3 description (p. 3) identifies archival and recall storage databases. The claim does not say these systems provide only basic storage. (LLM, conf=0.88)
- Supportedcites [39] [39] Wanjun Zhong, Lianghong Guo, Qiqi Gao, He Ye, and Yanlin Wang. Memorybank: Enhancing large language models with long-term memory. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 19724–19731, 2024. 12 Contents source
MemoryBank is an LLM memory system with a memory-storage component. Section 2 describes that component as its primary data repository, and §2.1 says it stores conversation records, event summaries, and user information. This supports citing MemoryBank as an example of a system that provides memory storage functionality. (LLM, conf=0.91)
- Not matchedcites [28] [28] Aymeric Roucher, Albert Villanova del Moral, Thomas Wolf, Leandro von Werra, and Erik Kaunismäki. 'smolagents': a smol library to build great agentic systems. https://github. com/huggingface/smolagents, 2025.
Its entry has no title and year that we can search by.
- Supportedcites [21] [21] Zhiwei Liu, Weiran Yao, Jianguo Zhang, Liangwei Yang, Zuxin Liu, Juntao Tan, Prafulla K Choubey, Tian Lan, Jason Wu, Huan Wang, et al. Agentlite: A lightweight library for building and advancing task-oriented llm agent system. arXiv preprint arXiv:2402.15538, 2024. source
Section 2.1, “Individual Agent” (page not available), describes AgentLite’s Memory component as storing an agent’s action-observation chains and saving new actions and observations during execution. This supports citing AgentLite as an example of an LLM-agent system with basic memory storage functionality. (LLM, conf=0.90)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“While previous datasets contain dialogues with around 1K tokens over 4-5 sessions, LoCoMo features much longer conversations averaging 9K tokens spanning up to 35 sessions”
§ 4 Experiment > 4.1 Dataset and Evaluation
What the sources say
- Supportedcites [22] [22] Adyasha Maharana, Dong-Ho Lee, Sergey Tulyakov, Mohit Bansal, Francesco Barbieri, and Yuwei Fang. Evaluating very long-term conversational memory of llm agents. arXiv preprint arXiv:2402.17753, 2024. source
The Introduction (p. 2) reports that LoCoMo conversations average 9K tokens and span up to 35 sessions. Its comparison table (p. 2) lists earlier multi-session datasets at about 1,055–1,226 tokens across four or five sessions, supporting the stated contrast. (LLM, conf=0.94)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“T-SNE Visualization of Memory Embeddings Showing More Organized Distribution with A-MEM (blue) Compared to Base Memory (red) Across Different Dialogues.”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“with each memory operation costing less than $0.0003 when using commercial API services”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
25 more claims
Create a free account to see all 63 claims and every source verdict.
- Probably not a substantial claim. 16
- It may state two separate claims as one. 20
- The quote only partly states this claim. 2
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source doesn’t support the claim as written.
In the citing paper
“Relative to the O (1) dictionary lookup used by all prior systems [4], [5], [6], [7], [8], [9], lineage-gated retrieval trades at most 13 . 82 µ s in absolute terms for the safety guarantee of Theorem 1-a tradeoff Experiments 1-2 show buys elimination of an 18 . 8 -25 . 5% structural leak rate at a 13 -14 percentage-point reuse cost.”
From Lineage-Aware Memory Governance: A Derivation-Gated Framework for Privacy-Preserving Column-Level Access Control in Enterprise AI Agents (2026) · cites this paper as [6]
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“This emerging class of systems extends the capabilities of traditional AI Agents by enabling multiple intelligent entities to collaboratively pursue goals through structured communication [144-146], shared memory [147,148], and dynamic role assignment [22].”
From AI Agents vs. Agentic AI: A Conceptual taxonomy, applications and challenges (2026) · cites this paper as [147]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“from Wang et al. [55], Xu et al. [45] build memory to improve long context performance for specific workflows which suggest that diverse evaluation workflows are needed for better testing of generalizable memory components.”
From MEMTRACK: Evaluating Long-Term Memory and State Tracking in Multi-Platform Dynamic Agent Environments (2025) · cites this paper as [45]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Some other approaches introduce external memory modules ( e.g., summarizers or retrievers) [63, 29, 13, 57], but these are typically trained separately and cannot be optimized end-to-end with the agent's policy. This also introduces additional engineering overhead, as engineers must manage and integrate two separate models.”
From MEM1: Learning to Synergize Memory and Reasoning for Efficient Long-Horizon Agents (2025) · cites this paper as [57]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Some others, such as HiAgent[5], A-Mem[23], MemAgent [28] use hierarchical designs to separate working buffers from episodic and long-term stores to relieve context pressure and improve persistence.”
From Learning Hierarchical Procedural Memory for LLM Agents through Bayesian Selection and Contrastive Refinement (2025) · cites this paper as [23]
Experimental — accuracy not yet measured
9 more claims about this paper
Create a free account to see every claim other papers make about this one.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ APPENDIX
“Using GPT-4o-mini, A-MEM achieves a ROUGE-L score of 44.27 in Multi-Hop tasks, more than doubling LoComo's 18.09.”
§ APPENDIX
“LOCOMO 9.64 23.92 2.01 18.09 3.40 11.58 26.48 40.20 60.46 69.59 READAGENT 2.47 9.45 0.95 13.12 0.55 5.76 2.99 9.92 6.66 9.79 MEMORYBANK 1.18 5.43 0.52 9.64 0.97 5.77 1.64 6.63 4.55 7.35 MEMGPT 10.58 25.60 4.76 25.22 0.76 9.14 28.44 42.24 36.62 43.75 A-MEM 10.61 25.86 21.39 44.27 3.42 12.09 29.50 45.18 42.62 50.04”
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ APPENDIX
“The Multi-Hop category showcases particularly striking results, where Qwen2.5-15b with A-MEM achieves a ROUGE-L score of 27.23, dramatically surpassing LoComo's 4.68 and ReadAgent's 2.81 - representing a nearly six-fold improvement.”
§ APPENDIX
“Multi Hop Temporal Open Domain Single Hop Adversial RGE-2 RGE-L RGE-2 RGE-L RGE-2 RGE-L RGE-2 RGE-L RGE-2 RGE-L”
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ APPENDIX
“This significant advantage maintains consistency across other metrics, with METEOR scores of 23.43 versus 7.61 and SBERT scores of 70.49 versus 52.30.”
§ APPENDIX
“Multi Hop Temporal Open Domain Single Hop Adversial ME SBERT ME SBERT ME SBERT ME SBERT ME SBERT”
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ 4 Experiment > 4.3 Empricial Results
“our A-MEM demonstrates superior performance in Multi-Hop tasks achieves at least two times better performance”
§ 4 Experiment > 4.1 Dataset and Evaluation
“MEMGPT 26.65 17.72 25.52 19.44 9.15 7.44 41.04 34.34 43.29 42.73 2.4 2.4 16,977 A-MEM 27.02 20.09 45.85 36.67 12.14 12.00 44.65 37.06 50.03 49.47 1.2 1.2 2,520”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 10 20 30 40 50
“44.55 25.43 28.31 32.12 34.32 37.02”
§ 4 Experiment > 4.1 Dataset and Evaluation
“A-MEM 27.02 20.09 45.85 36.67 12.14 12.00 44.65 37.06 50.03 49.47 1.2 1.2 2,520”
Citations
Cites
The publisher hasn’t deposited this paper’s reference list.
Research works printed but not linked: 4
- [1][1] Sönke Ahrens. How to Take Smart Notes: One Simple Technique to Boost Writing, Learning and Thinking. Amazon, 2017. Second Edition.
- [5][5] Satanjeev Banerjee and Alon Lavie. Meteor: An automatic metric for mt evaluation with improved correlation with human judgments. In Proceedings of the acl workshop on intrinsic and extrinsic evaluation measures for machine translation and/or summarization, pages 65–72, 2005.
- [18][18] Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, et al. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in Neural Information Processing Systems, 33:9459–9474, 2020.
- [19][19] Chin-Yew Lin. Rouge: A package for automatic evaluation of summaries. In Text summarization branches out, pages 74–81, 2004.
Cited by
Items 501–600 of 1,030
When Does Execution Provenance Help Agent Memory Retrieval?Preprint
Wang, Y., ju, J., Zhang, J., sun, Z., Duan, Y., Zheng, M., & Cai, T. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7476758 · Cited by 2
VibeMemBench: Evaluating Memory Systems for Coding Agents on Real Repository Coding TasksPreprint
Fan, L., Shi, Y., Li, Y., Sun, C., Chen, X., Xu, X., Wei, H., Ni, S., Yang, M., & Ye, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2609.23570 · Cited by 4
EnSIMem: Entity-Structured Indexing for Long-Term Agent MemoryPreprint
Meng, X., Fan, X., Fan, X., Guo, C., Xie, Y., & Han, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2609.27279 · Cited by 2
“EnSIMem complements these approaches with source-grounded entity-property access for factual and relational recall.”
Sangaraju, V., & Vissa, S. (2026) · IEEE Access · doi:10.1109/access.2026.3730363 · Cited by 1
“A-MEM [6] appeared at NeurIPS 2025.”
Hyperbolic Learning for Structured Data, Knowledge, and Memory: A Tutorial
Liu, J., Yang, M., & King, I. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3816467 · Cited by 1
“Recent work also opens opportunities in fully hyperbolic LLM pre-training, non-parametric memory, agent memory management, model editing, and unlearning [4, 18, 20, 26, 28, 35, 38, 46, 59, 63, 74, 82].”
InfluentialME-IQA: Memory-Enhanced Image Quality Assessment via Re-ranking
Fan, K., Wu, T., Wen, W., Liu, J., Yang, L., Zhang, Y., Liao, Y., Li, J., & Zhang, L. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37356-4_19 · Cited by 1
Self-citationSeek to Segment: Active Perception for Panoramic Referring Segmentation
Tang, S., Hu, S., Shuai, X., Ding, H., & Jiang, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37044-0_6 · Cited by 1
When Does Memory Help? A Cost-Aware Evaluation of Long-Term Memory in Tool-Using LLM Agents
Mishra, S., & Mishra, S. (2026) · arXiv · doi:10.48550/arxiv.2609.05441 · Cited by 1
Wu, C., Luo, Y., Qu, Z., & Wang, M. (2025) · arXiv · Cited by 1
“…or delete entries through reinforcement learning [3]; A-Mem, which adopts a Zettelkasten-inspired note-based architecture for autonomous retrieval [4]; and G-Memory, which introduces hierarchical graph-based structures to facilitate multi-agent interaction and improve embodied action success and…”
D-MEM: Dopamine-Gated Agentic Memory via Reward Prediction Error Routing
Song, Y., & Xin, Q. (2026) · arXiv · doi:10.48550/arxiv.2603.14597 · Cited by 1
“The state-of-the-art A-MEM framework [Weng et al., 2025] exemplifies this paradigm by structuralizing memory into a dynamic knowledge graph.”
Cherif, A. (2026) · arXiv · doi:10.48550/arxiv.2608.00009 · Cited by 1
“A-MEM [30] proposes an associative memory modelled on Zettelkasten note-taking, building inter-note links via LLM-generated keywords at the summary level.”
Hierarchical Memory for High-Efficiency Long-Term Reasoning in LLM Agents
Sun, H., & Zeng, S. (2025) · arXiv · doi:10.48550/arxiv.2507.22925 · Cited by 1
“As new memories are added, the system updates contextual representations of existing ones, progressively optimizing the knowledge structure and enhancing LLM agent performance in long-term interaction tasks (Xu et al., 2025).”
InfluentialEvo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory
Wei, T., Sachdeva, N., Coleman, B., He, Z., Bei, Y., Ning, X., Ai, M., Li, Y., He, J., Chi, E. H., et al. (2025) · arXiv · doi:10.48550/arxiv.2511.20857 · Cited by 1
Evaluating Memory in LLM Agents via Incremental Multi-Turn Interactions
Hu, Y., Wang, Y., & McAuley, J. (2025) · arXiv · doi:10.48550/arxiv.2507.05257 · Cited by 1
“3 MemoryAgentBench”
Westhäußer, R., Minker, W., & Zepf, S. (2026) · Proceedings of the 31st International Conference on Intelligent User Interfaces · doi:10.1145/3742413.3789222 · Cited by 1
“Therefore, we focused on MemoryBank and TiM to ensure a consistent comparison, as both use the same dataset and evaluation metrics.”
Grounding Agent Memory in Contextual Intent
Yang, R., Jiang, Y., Jiang, Y., Kargupta, P., Zhang, Y., & Han, J. (2026) · Findings of the Association for Computational Linguistics: ACL 2026 · doi:10.18653/v1/2026.findings-acl.584 · Cited by 1
Self-citationHingeMem: Boundary Guided Long-Term Memory with Query Adaptive Retrieval for Scalable Dialogues
Zhong, Y., Gao, Y., & Wang, H. (2026) · Proceedings of the ACM Web Conference 2026 · doi:10.1145/3774904.3792089 · Cited by 1
Externalization in LLM Agents: A Unified Review of Memory, Skills, Protocols and Harness Engineering
Zhou, C., Chai, H., Chen, W., Guo, Z., Shan, R., Song, Y., Xu, T., Yang, Y., Yu, A., Zhang, W., et al. (2026) · arXiv · doi:10.48550/arxiv.2604.08224 · Cited by 1
Meta Context Engineering via Agentic Skill Evolution
Ye, H., He, X., Arak, V., Dong, H., & Song, G. (2026) · arXiv · doi:10.48550/arxiv.2601.21557 · Cited by 1
ACON: Optimizing Context Compression for Long-horizon LLM Agents
Kang, M., Chen, W.-N., Han, D., Inan, H. A., Wutschitz, L., Chen, Y., Sim, R., & Rajmohan, S. (2025) · arXiv · doi:10.48550/arxiv.2510.00615 · Cited by 1
Rethinking Memory in LLM based Agents: Representations, Operations, and Emerging Topics
Du, Y., Huang, W., Zheng, D., Wang, Z., Montella, S., Lapata, M., Wong, K.-F., & Pan, J. Z. (2025) · arXiv · doi:10.48550/arxiv.2505.00675 · Cited by 1
TiMem: Temporal-Hierarchical Memory Consolidation for Long-Horizon Conversational Agents
Li, K., Yu, X., Ni, Z., Zeng, Y., Xu, Y., Zhang, Z., Li, X., Sang, J., Duan, X., Wang, X., et al. (2026) · arXiv · doi:10.48550/arxiv.2601.02845 · Cited by 1
Self-citationToolMem: Enhancing Multimodal Agents with Learnable Tool Capability Memory
Xiao, Y., Li, Y., Wang, H., Tang, Y., & Wang, Z. Z. (2025) · arXiv · doi:10.48550/arxiv.2510.06664 · Cited by 1
Memora: A Harmonic Memory Representation Balancing Abstraction and Specificity
Xia, M., Zhang, X., Dixit, S., Harimurugan, P., Wang, R., Ruhle, V., Sim, R., Bansal, C., & Rajmohan, S. (2026) · arXiv · doi:10.48550/arxiv.2602.03315 · Cited by 1
REMem: Reasoning with Episodic Memory in Language Agent
Shu, Y., Jonnalagedda, S. P., Gao, X., Gutiérrez, B. J., Qi, W., Das, K., Sun, H., & Su, Y. (2026) · arXiv · doi:10.48550/arxiv.2602.13530 · Cited by 1
APEX-MEM: Agentic Semi-Structured Memory with Temporal Reasoning for Long-Term Conversational AI
Banerjee, P., Moshtaghi, M., Subramanian, S., Misra, A., & Chadha, A. (2026) · Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2026.acl-long.749 · Cited by 1
Thinking on maps: how foundation model agents explore, remember, and reason in map environments
Wei, Z., Liu, Y., Liao, H., & Xu, W. (2026) · GIScience & Remote Sensing · doi:10.1080/15481603.2026.2730755 · Cited by 1
Self-citationTang, Z., Ji, B., Sun, R., Wang, H., Li, J., & Yijun, Z. (2026) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/2026.emnlp-main.2574 · Cited by 2
Dai, R., Hu, H., Zhang, J., Kang, Y., & Tan, A. (2026) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v40i25.39205 · Cited by 3
Task-Aligned Tool Recommendation for Large Language Models
Gao, H., & Zhang, Y. (2025) · International Joint Conference on Natural Language Processing and Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (IJCNLP-AACL) · doi:10.18653/v1/2025.ijcnlp-long.110 · Cited by 4
“Memory recommendation (Borgeaud et al., 2022; Gao and Zhang, 2024a; Xu et al., 2025) facilitates the retrieval of relevant past experiences or interactions, improving continuity, consistency, and long-term context in multi-turn conversations.”
Self-citationDiffBench Meets DiffAgent: End-to-End LLM-Driven Diffusion Acceleration Code Generation
Jiao, J., Zhu, H., Yang, P., Wang, J., Liu, J., Liu, Z., Li, D., Fang, Y., Yong, J., Wang, B., et al. (2026) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v40i27.39395 · Cited by 1
Banerjee, P., Moshtaghi, M., & Chadha, A. (2026) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/2026.findings-emnlp.1996 · Cited by 1
LBM: Hierarchical Large Auto-Bidding Model via Reasoning and Acting
Li, Y., Lyu, Z., Jiang, P., Cai, Q., Pan, F., An, B., & Jiang, P. (2026) · International World Wide Web Conference (WWW) · doi:10.1145/3774904.3792202 · Cited by 4
“Early studies have proposed structured workflows that integrate reasoning, action, and reflection steps such as ReAct [51] and Reflexion [41], as well as interaction through memory [50], retrieval systems [14], and external tools [43].”
Metaphors for Memory: Charting a Design Space of AI Memory Tools and Interfaces
Kim, M. G., Famelis, M., & Arawjo, I. (2026) · Designing Interactive Systems Conference · doi:10.1145/3800645.3812979 · Cited by 2
MemAura: Structured Context Memory for Personalized LLM Reasoning in Smart Environments
Liu, S., & Chen, H. (2026) · Proceedings of the ACM on Interactive Mobile Wearable and Ubiquitous Technologies · doi:10.1145/3810196 · Cited by 2
SlideSAVR: Enabling Live Analysis during Data Presentations via Multimodal Sketching and Voice Input
Han, C., Tanjim, M. M., Guo, S., Dierk, C., Isaacs, K. E., & Hoffswell, J. (2026) · Computer Graphics Forum · doi:10.1111/cgf.70459 · Cited by 2
A Survey of Large-Model-Based AI Agents
Qin, C., & Jin, L. (2026) · Tsinghua Science & Technology · doi:10.26599/tst.2026.9010072 · Cited by 2
PerMemSafe: Benchmarking Implicit Personalized Safety of Long Horizon Self-Evolving Agents
An, H., Li, M., Xu, N., Zhou, C., Xu, X., Du, T., Li, J., & Ji, S. (2026) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2026.findings-acl.320 · Cited by 1
RealMem: Benchmarking LLMs in Real-World Memory-Driven Interaction
Bian, H., Yao, Z., Hu, S., Xu, Z., Zhang, S., Guo, Y., Yang, Z., Han, X., Wang, H., & Chen, R. (2026) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2026.findings-acl.703 · Cited by 6
U-Fold: Dynamic Intent-Aware Context Folding for User-Centric Agents
Jin, S., Fang, R., Li, Y., Wang, X., Cai, S., Xie, P., Zhang, N., & Yuan, F. (2026) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2026.findings-acl.897 · Cited by 1
Agent-Dice: Disentangling Knowledge Updates via Geometric Consensus for Agent Continual Learning
Wu, Z., Lou, X., Ma, X., 李燕思, Liu, W., Zhang, W., Wang, J., & Zhang, Z. (2026) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2026.findings-acl.908 · Cited by 1
HiGMem: A Hierarchical and LLM-Guided Memory System for Long-Term Conversational Agents
Cao, S., He, J., & Tan, F. (2026) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2026.findings-acl.1690 · Cited by 1
Chen, M.-B., Lau, J. H., & Frermann, L. (2026) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2026.acl-long.2203 · Cited by 2
MemoPhishAgent: Memory-Augmented Multi-Modal LLM Agent for Phishing URL Detection
Chen, X., Liu, H., Yuan, T., Kafai, M., Habas, P., & Zhang, X. (2026) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2026.acl-industry.84 · Cited by 3
ClinTraceBench: Source-Verifiable Longitudinal Clinical Reasoning over EHR-Derived Dialogues
Wang, H., Zhao, Z., & Zhao, Y. (2026) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/2026.findings-emnlp.1480 · Cited by 1
SAGE-QA: Shared-context agentic graph-enhanced question answering for scientific reasoning
Hsu, Y.-C., & Buehler, M. J. (2026) · MRS Communications · doi:10.1557/s43579-026-01036-4 · Cited by 1
Awais, M., & Ayaz, M. (2026) · International Journal of Engineering Technology Research & Management · doi:10.67809/ijetrm.1789357569 · Cited by 1
Agent Memory with Episodic Retrieval for Financial Decision-Making
Xu, N., Liu, J., Huang, W., Zhang, X., Cao, J., & Wei, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2609.28771 · Cited by 1
“While dynamic memory frameworks such as Mem0 and MemGPT (Chhikara et al., 2025; Packer et al., 2023; Xu et al., 2025) enable long-term reasoning in text-based tasks, they fail to generalize to the stochastic, high-dimensional structure of financial time series.”
RCR-Router:Role-Conditioned Context Routing for Resource-Efficient Multi-Agent LLM Systems
Liu, J., Zhao, P., Kong, Z., Zhang, W., Yang, F., Dong, P., Yang, C., Fan, W., Li, T., Tang, H., et al. (2026) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/2026.findings-emnlp.425 · Cited by 2
Ouyang, L. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2507.21055 · Cited by 1
“Xu et al. [16] introduced ’A-mem: Agentic Memory for LLM agents,’ highlighting the importance of sophisticated memory structures for more human-like understanding and reasoning in LLMs.”
$How^{2}$: How to learn from procedural How-to questionsPreprint
Dagan, G., Keller, F., & Lascarides, A. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2510.11144 · Cited by 1
“A third line of work investigates knowledge organisation over time, using hierarchical memory (Packer et al., 2024), knowledge networks (Xu et al., 2025), or structured rule libraries (Chen et al., 2024).”
Learning to Debug: LLM-Organized Knowledge Trees for Solving RTL Assertion FailuresPreprint
Bai, Y., & Ren, H. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2511.17833 · Cited by 2
“These stores can be dynamic graphs [49], executable skill libraries [41], or architectural augmentations [23], [43].”
Data in Agentic AI: A Comprehensive SurveyPreprint
Jin, Y., Wang, H., Zhang, Y., Wen, D., Qin, L., & Zhang, W. (2025) · doi:10.36227/techrxiv.176704339.93520117/v1 · Cited by 1
Self-citationImprove Large Language Model Systems with User LogsPreprint
Wang, C., Su, W., Ai, Q., Yue, X., Zhang, R., Chang, X., & Liu, Y. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2602.06470 · Cited by 1
Closing Reasoning Gaps in Clinical Agents with Differential Reasoning LearningPreprint
Liu, J., Jiang, Y., Krishnan, R., Padman, R., Zhang, Y., & Bian, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2602.09945 · Cited by 1
“Because clinical questions are typically multi-factorial involving findings, intermediate hypotheses, and recommended actions, we represent each reasoning trace as a reasoning graph: a directed acyclic graph (DAG) that makes the underlying relational structure explicit.”
Self-citationAMV-L: Lifecycle-Managed Agent Memory for Tail-Latency Control in Long-Running LLM SystemsPreprint
Bamidele, E. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.04443 · Cited by 2
CLAG: Adaptive Memory Organization via Agent-Driven Clustering for Small Language Model AgentsPreprint
Roh, T., Jang, W., Jung, J., & Kang, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.15421 · Cited by 1
“We compare our framework against a standard RAG (Lewis et al., 2020) baseline and representative agentic memory architectures including A-mem (Xu et al., 2025), MemoryOS (Kang et al., 2025), and GAM (Yan et al., 2025); detailed implementation settings are provided in Appendix B.”
InfluentialMemory as Asset: From Agent-centric to Human-centric Memory ManagementPreprint
Pan, Y., Huang, Q., & Yang, W. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.14212 · Cited by 1
CraniMem: Cranial Inspired Gated and Bounded Memory for Agentic SystemsPreprint
Mody, P., Panchal, M., kar, R., Bhowmick, K., & Karani, R. B. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.15642 · Cited by 1
“A-Mem formalizes agentic memory as a component that supports writing experiences and retrieving them for downstream planning and decision making (Xu et al., 2025).”
Breaking the Capability Ceiling of LLM Post-Training by Reintroducing Markov StatesPreprint
Yuan, Y., & Xie, T. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.19987 · Cited by 1
View-oriented Conversation Compiler for Agent Trace AnalysisPreprint
Zhang, L., & Agrawala, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.29678 · Cited by 1
Cognis: Context-Aware Memory for Conversational AI AgentsPreprint
Daftari, P., Patel, K., Kapale, S., George, J., & Surendira, S. (2026) · arXiv (Cornell University) · doi:10.13140/rg.2.2.28445.27365 · Cited by 1
Memory as Metabolism: A Design for Companion Knowledge SystemsPreprint
Miteski, S. (2026) · arXiv.org · doi:10.48550/arxiv.2604.12034
Harnessing Pre-Resolution Signals for Future Prediction AgentsPreprint
Wei, C., Gao, M., Han, Z., Chen, K., Zhuang, Y., Guan, H., Zhang, Y., Cheng, Y., Zhou, X., Chen, H., et al. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.15719 · Cited by 1
“…experience into reusable natural-language lessons, Voyager builds an explicit skill library, GEPA evolves prompts through reflection, and A-MEM studies long-term memory organization for LLM agents [Shinn et al., 2023, Zhao et al., 2024, Wang et al., 2023, Agrawal et al., 2025, Xu et al., 2025].”
Self-citationGRAB-ANNS: High-Throughput Indexing and Hybrid Search via GPU-Native BucketingPreprint
Zhao, X., Lou, H., Zhang, Y., Dai, J., Deng, S., & Yin, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.16402 · Cited by 1
Self-citationTo Know is to Construct: Schema-Constrained Generation for Agent MemoryPreprint
Zheng, L., Song, W., Li, D., & Yang, Y. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.20117 · Cited by 1
“Since semantic similarity is not equivalent to contextual relevance, embeddings often fail to distinguish semantically identical but contextually distinct instances [ Xu et al. , 2025a ] .”
FlowEvo: Self-Evolving Agents through the Co-Evolution of Workflows and Executable SkillsPreprint
Ren, Z., Yue, L., Li, R., Wang, Y., Xu, S., Liu, H., Pan, S., & Di, S. (2026) · arXiv (Cornell University) · doi:10.13140/rg.2.2.13642.32960 · Cited by 4
Web2BigTable: A Bi-Level Multi-Agent LLM System for Internet-Scale Information Search and ExtractionPreprint
Huang, Y., Chen, Y., (1836730), Z. H., Chen, Y., Lee, K. Y., Zhou, H., Luo, W., Fang, M., & Wang, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.27221 · Cited by 1
“…recent self-evolving and memory-augmented agents improve execution through reusable external skills [7, 20, 17] or structured long-term memory [23, 24], yet this adaptation has largely been studied at a single level, without jointly refining how tasks are decomposed and how sub-tasks are…”
Reliable AI Needs to Externalize Implicit Knowledge: A Human-AI Collaboration PerspectivePreprint
Liu, H., Li, T., Cui, Z., Li, Y., Wu, Z., Pedersen, T. B., Torp, K. M., & Jensen, C. S. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.02010 · Cited by 1
“MemoryBank (Zhong et al., 2024) models forgetting using Ebbinghaus decay curves, while A-MEM (Xu et al., 2025) dynamically organizes knowledge into interconnected networks.”
Empowering VLMs for Few-Shot Multimodal Time Series Classification via Tailored Agentic Reasoning
Li, L., Huang, J., Quan, Q., Li, D., Li, B., Zhang, X., Meng, E., Feng, W., Lou, J., & Ng, S.-K. (2026) · ACM International Conference on Multimedia (ACM MM) · doi:10.48550/arxiv.2605.09395 · Cited by 1
Nautilus Compass: Black-box Persona Drift Detection for Production LLM AgentsPreprint
Wang, C. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.09863 · Cited by 4
Omni-Persona: Systematic Benchmarking and Improving Omnimodal PersonalizationPreprint
Oh, Y., Lee, D., Park, S., Kim, H., & Yoon, S. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.09996 · Cited by 1
“With approximately 50% of the evaluation samples being no-GT, the benchmark systematically probes the model’s resistance to hallucination, an essential desideratum when integrating with RAG systems [19–21].”
Context Training with Active Information SeekingPreprint
Huang, Z., Kuncoro, A., Feng, Q., Shen, J., Dery, L. M., Szlam, A., & Ranzato, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.13050 · Cited by 1
“Leveraging the effectiveness of In-Context Learning (ICL) in Large Language Models (LLMs), recent advancements in context engineering aim to transform static context into a dynamic working memory (Packer et al., 2023; Xu et al., 2025b).”
Evaluating Temporal Semantic Caching and Workflow Optimization in Agentic Plan-Execute PipelinesPreprint
Merchant, A. M., Veera, K., Goyla, S. K., Bhure, S., Patel, D., & Maghraoui, K. E. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.20630 · Cited by 1
“Several recent systems augment LLM agents with external memory (14; 15; 16).”
Rethinking Memory as Continuously Evolving ConnectivityPreprint
Fang, J., BuQiang, X., Wang, Z., 曹浩亮, Deng, X., Dong, B., Zhu, H., 黃若惠, Yu, G., Wei, Y., et al. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.28773 · Cited by 3
HEART-Bench: Do LLMs Make Persona-Consistent Decisions?Preprint
Peng, W., Zhang, C., Wang, Q., Shi, Y., Lian, H., Mao, Q., Pang, J., Feng, C., Li, B., & Gu, X. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.30058 · Cited by 1
Absorbing Complexity: An Interaction-Native Knowledge Harness for Financial LLM AgentsPreprint
Borjigin, A., Stadnyk, I., Bilski, B., Chikita, M., Kyrylenko, D., Pidturkina, S., & Stadnyk, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.01886 · Cited by 1
“A-MEM pushes further toward dynamically organized “agentic memory” [18].”
SaliMory: Orchestrating Cognitive Memory for Conversational AgentsPreprint
Zhang, K., Zhang, X., Jiang, H., Kuo, S.-Z., Yun, H., Ahmed, E., Oraby, S., Li, Z., Sharma, S., Lee, A., et al. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.04120 · Cited by 1
“Another trend is to actively compress and curate long-term memories, but neither prompting-based (Li et al., 2024; Xu et al., 2025b) nor reinforcement-learning–based (Yan et al., 2025; Wang et al., 2025) variants have proved to maintain high-quality memory, which reliably captures user preferences…”
When More Cores Hurts: The Vector Database Scaling Paradox in HPCPreprint
Ockerman, S., Oh, S. Y., Gueroudji, A., Chaturvedi, R., Carns, P., Chia, N., Dorier, M., Latham, R., Mallick, T., Perarnau, S., et al. (2026) · arXiv (Cornell University) · doi:10.3929/ethz-c-000806110 · Cited by 1
MetaEvo: A Meta-Optimization Framework for Experience-Driven Agent EvolutionPreprint
Ren, B., Huang, H., Li, Y., & Gao, Y. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.07603 · Cited by 1
“Beyond static storage, A-MEM (Xu et al., 2025b) and Mem0 (Chhikara et al., 2025) explicitly model memory evolution, enabling agents to refine and reorganize stored knowledge over time.”
Functional Cache Grafting: Robust and Rapid Code-Policy Synthesis for Embodied AgentsPreprint
Chun, S., Choi, W., Choi, S., Ahn, S., & Woo, H. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.13097 · Cited by 1
“Recent work has explored equipping embodied agents with memory to support long-horizon reasoning, task adaptation, and generalization by reusing validated experiences (Xu et al., 2025; Shinn et al., 2024; Li et al., 2025b; Kang et al., 2023; Wang et al., 2025; Kagaya et al., 2024).”
Marginal Advantage Accumulation for Memory-Driven Agent Self-EvolutionPreprint
Yang, M., Zheng, K., Cheng, C., Liu, Y., Lu, X., Jiang, F., & Zheng, Y. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.20475 · Cited by 2
Ding, H., Yu, X., Wang, C., Xiao, J., Bao, K., Wang, W., & He, X. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.23283 · Cited by 2
“To extract relevant knowledge from large memory banks, recent LLM agents employ lexical [27, 37, 42], semantic [8, 28], graph-based [12, 45, 47], and hybrid retrieval [1, 2, 41], often enhanced by query reconstruction (e.g., decomposition or rewriting) [14], re-ranking [7, 14, 49], or compression…”
Turning Intent into Specifications: A Benchmark and an Interactive User-Assistant AgentPreprint
Wang, H., Han, L., Xu, K., & Srivastava, A. K. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.20585 · Cited by 1
“For example, [XLM + 25] propose A-Mem that autonomously builds and (has everything the user needs) • Precision: (avoids extra or incorrect details) • Consistency: (internally correct and coherent) • Insight: (ideas the user didn’t think of) • Readability: (easy to understand) ≤ T rounds Figure 1:…”
Goel, S., Vaidhyanathan, P., Schorling, L., Ares, N., & Osborne, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.20638 · Cited by 1
Libra: Training the Environment for Agentic Information RetrievalPreprint
Zhao, X., Chiu, A., & Wang, G. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2607.00016 · Cited by 1
ReflectWorld-MM: An Entity-Oriented Multimodal Memory System for Open-Ended Video StreamsPreprint
Ma, X., Sun, Y., Jin, Z., Gu, J., Luo, Y., Shao, S., Tang, C., Chen, J., & Pu, L. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2607.09759 · Cited by 1
Beyond Memory Leaderboards: Evaluating Scientific Memory as Budgeted Context RestorationPreprint
Sheverev, M., Finkelstein, D., & Nikolenko, S. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2607.16848 · Cited by 1
“A-MEM [46] casts each memory as a Zettelkasten note (raw content, timestamp, LLM-extracted keywords and tags, a generated contextual description, an embedding, and links); when a new note is added, the system links it to its nearest neighbors and, crucially, rewrites the descriptions and tags of…”
Supra Cognitive Modes: A Routed Architecture for Agent MemoryPreprint
Tobkin, J., & Yang, D. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2607.19096 · Cited by 1
“Mem0 extracts persistent user-specific memories at ingest [4]; Zep/Graphiti uses a temporal knowledge graph with provenance [19]; A-Mem organizes memories as dynamically linked notes [30]; and MIRIX uses several typed stores under a controller [27].”
Knowledge-Centric Self-ImprovementPreprint
Wang, X. J., Yoon, L. H., Qu, C., Wang, A. F., Sehgal, A., Mazumdar, E., & Yue, Y. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2607.19592 · Cited by 2
“Building on retrieval augmentation [17, 5], many systems study how to store, retrieve, link, and maintain memories so that agents can better leverage prior context [21, 49, 34, 10].”
Delivery, Not Storage: Cue-Anchored Working Memory as a Harness Property for Coding AgentsPreprint
Saha, S. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2607.20972 · Cited by 1
SF-AMS: Strategic Forgetting for Structured Memory in LLM AgentPreprint
Yang, N., Li, S., Shen, M., Zhou, Y., Zhang, M., Li, T., & Zhang, H. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2607.22562 · Cited by 1
“Knowledge organization methods aim to improve memory quality by transforming raw interactions into structured and interpretable representations (Xu et al., 2025; Chhikara et al., 2025; Modarressi et al., 2023).”
InfluentialSkillSmith: Learning to Compose Parametric Skills and Textual KnowledgePreprint
Dery, L. M., Tjandra, B. A., Samiei, S., Kuncoro, A., Yahav, Z., Shen, J., & Szlam, A. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2607.27497 · Cited by 1
“…where agents use natural language—such as self-reflection (Madaan et al., 2023; Renze and Guven, 2024; Shinn et al., 2023), structured memory (Chhikara et al., 2025; Xu et al., 2025a,b), or prompt generation (Agrawal et al., 2026; Wang et al., 2024)—to guide future reasoning and planning.”
Personalizing Large Language Model Agents with Small Policy ModelsPreprint
Jin, D., Zhang, Z., Li, H., Pan, Y., Huang, R., & Zhou, D. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.00215 · Cited by 1
“LaMP benchmarks profile-conditioned personalization (Salemi et al. 2024), while MemoryBank, MemGPT, A-MEM, and Mem0 develop mechanisms for storing and retrieving information across interactions (Zhong et al. 2023; Packer et al. 2023; Xu et al. 2025; Chhikara et al. 2025).”
TrajWiki: Source-Grounded Memory Trajectories for Long-Horizon Dialogue AgentsPreprint
Sun, J., Xue, Y., Li, M., Yao, Z., Li, J., Cui, Y., Cai, W., Liu, H., Wang, F., Montgomery, M. K., et al. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.00967 · Cited by 4
“Across both benchmarks and both GPT-and Qwen-based backbones, TrajWiki consistently outperforms most of the baselines, including Full Context, Naive RAG, LangMem, A-MEM, and Mem0 [21, 22, 35].”
InfluentialArgus: A General-Purpose Agentic Reasoning Runtime for Long-Horizon TasksPreprint
Li, B., Wen, Z., Fan, Y., Wen, C., Yang, F., Guo, H., Wu, J., Zhang, J., Lei, J., Li, M., et al. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.05144 · Cited by 1
“This follows the broader move from monolithic context toward managed memory tiers and reusable workflows (Packer et al., 2023; Wang et al., 2024b; Xu et al., 2025).”
Self-citationCoupling Planning with Episodic Memory in LLM Agents for Software Issue ResolutionPreprint
Zhang, J., Zhang, Y., & Huang, Y. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.06811 · Cited by 1
“Memory-centric work expands what the agent can retrieve: reflective and long-horizon stores [4], [40]–[43], and, for issue resolution, experience memories distilled from past attempts [8]–[10]; planning there stays implicit or pipeline-fixed.”
Self-citationSodaMem: Evidence-Grounded Temporal Graph Memory for LLM AgentsPreprint
Wan, F., Wu, C., & Lyu, N. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.08055 · Cited by 2
“…work spans a familiar pipeline—land, structure, index, link, maintain, retrieve/answer (Agent-Native 2026; Huang et al. 2026; Lewis et al. 2020; Xu et al. 2025; Ras-mussen et al. 2025; Liu et al. 2026; Memory-R1 2025; Nap-Mem 2026; Chhikara et al. 2025)—but everyday assistants still hit four…”
Total Recall at What Cost? Benchmarking the Serving Cost of Agentic Memory SystemsPreprint
Pollertlam, N., & Kornsuwannawit, W. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.11879 · Cited by 1
“Later systems pursue richer representations, including interlinked memory notes [30], temporally-aware knowledge graphs [24], and durative summaries of temporally continuous facts [27].”
D$^2$ACCI: A Dual-Loop Diagnostic Protocol for Evidence-Preserving Agent MemoryPreprint
Liu, X., Liu, Y., Li, C., & Kun, S. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.17756 · Cited by 1
Metadata
- Authors
- Xu, W., Liang, Z., Mei, K., Gao, H., Tan, J., & Zhang, Y.
- Affiliations
- Rutgers, The State University of New Jersey (US)
- Ant Group (China) (CN)
- Salesforce (United States) (US)
- Venue
- Advances in Neural Information Processing Systems 38
- Published
- 2025
- Pages
- 20004-20031
- ISSN
- 1049-5258 (ISSN-L)
- OpenAlex
- W4407759090
- Cited by
- 51
- References
- 0
- Citations by year
- 2026: 48 · 2025: 3
- FWCI
- 48.02
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- neural information processing systems (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › AI-based Problem Solving and Planning
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Information Systems › Information Retrieval and Search Behavior
- Keywords
- agentic memory, LLM agents, knowledge networks, memory organization, dynamic indexing, evolution of memory, knowledge representation, memory management
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-28, under version 3 of the claim definition (prompt extract-v3).
A second model labelled each claim on 2026-09-28 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.