Data Management for Agentic Memory: Foundations, Systems, and Challenges

Contents

• Data Management for Agentic Memory: Foundations, Systems, and Challenges
  • Contents
  • Introduction
  • Paper List
    • Memory Categorization
      • Short-Term Memory
      • Long-Term Memory
        • Episodic Memory
        • Semantic Memory
        • Procedural Memory
        • Personal Memory
    • Memory Management
      • Operators
        • Extract
        • Update
        • Conflict Detection
        • Retrieval
        • Consolidation
        • Forgetting
        • Compression
        • Reflection
      • Safety Management
        • Access Control
        • Privacy Protection
    • Memory Storage
      • Textual Memory
        • Structured Memory
        • Semi-Structured Memory
        • Unstructured Memory
      • Model-Level Memory
        • Parameter Storage
        • Latent Storage
    • Survey
  • Open Problems

Introduction

Recent advances in large language models (LLMs) have driven the rapid emergence of autonomous agents as a transformative paradigm for building intelligent AI systems. By integrating reasoning, planning, tool use, and interaction capabilities, these agents have shown immense potential in addressing complex, open-ended tasks. However, despite their growing sophistication, most existing agents remain fundamentally stateless -- processing each task or interaction in isolation, with little ability to retain, organize, or reuse past experiences over time. This lack of memory severely limits their ability to personalize, adapt over the long term, and engage in lifelong learning. Without mechanisms to accumulate experiences, reflect on past failures, or refine behavior across interactions, their potential for true autonomy remains constrained. To overcome this limitation, agentic memory has emerged as a critical area of research. Agentic memory equips agents with the ability to store, retrieve, update, and utilize information from prior interactions, internal reasoning trajectories, tool executions, and environmental feedback. This enables agents to build a dynamic repository of knowledge, moving far beyond simple context buffers or append-only logs.

From a systems perspective, agentic memory shares striking parallels with database systems. Both are responsible for storing, organizing, retrieving, updating, and maintaining information to support reasoning and decision-making in dynamic environments. However, integrating agentic memory with database systems introduces a transformative opportunity. While databases excel at storing and indexing structured data, the rise of LLMs and semantic analytics demands more flexible and dynamic memory management. Agentic memory complements databases by enhancing query functionality -- enabling natural language queries with contextual memory, dynamic query plan refinement, and semantic reasoning across structured and unstructured data sources. Additionally, it improves execution efficiency by leveraging historical insights to optimize query pipelines and reduce computation costs.

The integration of agentic memory and databases not only amplifies the capabilities of autonomous agents but also redefines the role of databases in adaptive, intelligent data systems. This synergy highlights a pressing need for the database community to advance agentic memory techniques. In this tutorial, we introduce the foundational techniques of agentic memory and aim to inspire further innovation at this exciting frontier. We systematically examine agentic memory systems from a data management perspective, as illustrated in the figure above, with the goal of providing participants with a comprehensive understanding of the field and highlighting key research challenges for advancing agentic memory capabilities.

Paper List

Memory Categorization

Short-Term Memory

1. LightMem: Lightweight and Efficient Memory-Augmented Generation Jizhan Fang, Xinle Deng, Haoming Xu, et al. CoRR 2025. [Paper]
2. Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation Zhanghao Hu, Qinglin Zhu, Hanqi Yan, et al. CoRR 2026. [Paper]
3. MemOS: A Memory OS for AI System Zhiyu Li, Shichao Song, Chenyang Xi, et al. CoRR 2025. [Paper]
4. MemGPT: Towards LLMs as Operating Systems Charles Packer, Vivian Fang, Shishir G. Patil, et al. CoRR 2023. [Paper]
5. Memory OS of AI Agent Jiazheng Kang, Mingming Ji, Zhe Zhao, et al. EMNLP 2025. [Paper]
6. MemoBrain: Executive Memory as an Agentic Brain for Reasoning. Hongjin Qian, Zhao Cao, Zheng Liu. ACL 2026. [Paper]
7. MemoryBank: Enhancing Large Language Models with Long-Term Memory Wanjun Zhong, Lianghong Guo, Qiqi Gao, et al. AAAI 2024. [Paper]

Long-Term Memory

Episodic Memory

1. AriGraph: Learning Knowledge Graph World Models with Episodic Memory for LLM Agents Petr Anokhin, Nikita Semenov, Artyom Y. Sorokin, et al. IJCAI 2025. [Paper]
2. Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation Zhanghao Hu, Qinglin Zhu, Hanqi Yan, et al. CoRR 2026. [Paper]
3. MemoryBank: Enhancing Large Language Models with Long-Term Memory Wanjun Zhong, Lianghong Guo, Qiqi Gao, et al. AAAI 2024. [Paper]

Semantic Memory

1. Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory Prateek Chhikara, Dev Khant, Saket Aryan, et al. ECAI 2025. [Paper]
2. Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation Zhanghao Hu, Qinglin Zhu, Hanqi Yan, et al. CoRR 2026. [Paper]

Procedural Memory

1. HiAgent: Hierarchical Working Memory Management for Solving Long-Horizon Agent Tasks with Large Language Model Mengkang Hu, Tianxing Chen, Qiguang Chen, et al. ACL 2025. [Paper]
2. Agent Workflow Memory Zora Zhiruo Wang, Jiayuan Mao, Daniel Fried, et al. ICML 2025. [Paper]

Personal Memory

1. Memory OS of AI Agent Jiazheng Kang, Mingming Ji, Zhe Zhao, et al. EMNLP 2025. [Paper]
2. MemoryBank: Enhancing Large Language Models with Long-Term Memory Wanjun Zhong, Lianghong Guo, Qiqi Gao, et al. AAAI 2024. [Paper]

Memory Management

Operators

Extract

1. LightMem: Lightweight and Efficient Memory-Augmented Generation Jizhan Fang, Xinle Deng, Haoming Xu, et al. CoRR 2025. [Paper]
2. Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory Prateek Chhikara, Dev Khant, Saket Aryan, et al. ECAI 2025. [Paper]
3. MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents Dongming Jiang, Yi Li, Guanpeng Li, Bingzhe Li. ACL 2026. [Paper]
4. MemLLM: Finetuning LLMs to Use An Explicit Read-Write Memory Ali Modarressi, Abdullatif Köksal, Ayyoob Imani, et al. CoRR 2024. [Paper]
5. Memory OS of AI Agent Jiazheng Kang, Mingming Ji, Zhe Zhao, et al. EMNLP 2025. [Paper]
6. Collaborative Memory: Multi-User Memory Sharing in LLM Agents with Dynamic Access Control Alireza Rezazadeh, Zichao Li, Ange Lou, et al. CoRR 2025. [Paper]
7. Agent Workflow Memory Zora Zhiruo Wang, Jiayuan Mao, Daniel Fried, et al. ICML 2025. [Paper]

Update

1. Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory Prateek Chhikara, Dev Khant, Saket Aryan, et al. ECAI 2025. [Paper]
2. ChatDB: Augmenting LLMs with Databases as Their Symbolic Memory Chenxu Hu, Jie Fu, Chenzhuang Du, et al. CoRR 2023. [Paper]
3. MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents Dongming Jiang, Yi Li, Guanpeng Li, Bingzhe Li. ACL 2026. [Paper]
4. From Isolated Conversations to Hierarchical Schemas: Dynamic Tree Memory Representation for LLMs Alireza Rezazadeh, Zichao Li, Wei Wei,et al. ICLR 2025. [Paper]
5. Memory-R1: Enhancing Large Language Model Agents to Manage and Utilize Memories via Reinforcement Learning Sikuan Yan, Xiufeng Yang, Zuchao Huang, et al. ACL 2026. [Paper]

Conflict Detection

1. Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory Prateek Chhikara, Dev Khant, Saket Aryan, et al. ECAI 2025. [Paper]
2. ChatDB: Augmenting LLMs with Databases as Their Symbolic Memory Chenxu Hu, Jie Fu, Chenzhuang Du, et al. CoRR 2023. [Paper]
3. Zep: A Temporal Knowledge Graph Architecture for Agent Memory Preston Rasmussen, Pavlo Paliychuk, Travis Beauvais, et al. CoRR 2025. [Paper]
4. Memory-R1: Enhancing Large Language Model Agents to Manage and Utilize Memories via Reinforcement Learning Sikuan Yan, Xiufeng Yang, Zuchao Huang, et al. ACL 2026. [Paper]

Retrieval

1. AriGraph: Learning Knowledge Graph World Models with Episodic Memory for LLM Agents Petr Anokhin, Nikita Semenov, Artyom Y. Sorokin, et al. IJCAI 2025. [Paper]
2. ChatDB: Augmenting LLMs with Databases as Their Symbolic Memory Chenxu Hu, Jie Fu, Chenzhuang Du, et al. CoRR 2023. [Paper]
3. Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation Zhanghao Hu, Qinglin Zhu, Hanqi Yan, et al. CoRR 2026. [Paper]
4. MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents Dongming Jiang, Yi Li, Guanpeng Li, Bingzhe Li. ACL 2026. [Paper]
5. MemLLM: Finetuning LLMs to Use An Explicit Read-Write Memory Ali Modarressi, Abdullatif Köksal, Ayyoob Imani, et al. CoRR 2024. [Paper]
6. Zep: A Temporal Knowledge Graph Architecture for Agent Memory Preston Rasmussen, Pavlo Paliychuk, Travis Beauvais, et al. CoRR 2025. [Paper]

Consolidation

1. Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation Zhanghao Hu, Qinglin Zhu, Hanqi Yan, et al. CoRR 2026. [Paper]
2. MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents Dongming Jiang, Yi Li, Guanpeng Li, Bingzhe Li. ACL 2026. [Paper]
3. Memory OS of AI Agent Jiazheng Kang, Mingming Ji, Zhe Zhao, et al. EMNLP 2025. [Paper]
4. Zep: A Temporal Knowledge Graph Architecture for Agent Memory Preston Rasmussen, Pavlo Paliychuk, Travis Beauvais, et al. CoRR 2025. [Paper]

Forgetting

1. MemOS: A Memory OS for AI System Zhiyu Li, Shichao Song, Chenyang Xi, et al. CoRR 2025. [Paper]
2. MemGPT: Towards LLMs as Operating Systems Charles Packer, Vivian Fang, Shishir G. Patil, et al. CoRR 2023. [Paper]
3. Memory OS of AI Agent Jiazheng Kang, Mingming Ji, Zhe Zhao, et al. EMNLP 2025. [Paper]
4. MemoryBank: Enhancing Large Language Models with Long-Term Memory Wanjun Zhong, Lianghong Guo, Qiqi Gao, et al. AAAI 2024. [Paper]

Compression

1. LightMem: Lightweight and Efficient Memory-Augmented Generation Jizhan Fang, Xinle Deng, Haoming Xu, et al. CoRR 2025. [Paper]

Reflection

1. HiAgent: Hierarchical Working Memory Management for Solving Long-Horizon Agent Tasks with Large Language Model Mengkang Hu, Tianxing Chen, Qiguang Chen, et al. ACL 2025. [Paper]
2. Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation Zhanghao Hu, Qinglin Zhu, Hanqi Yan, et al. CoRR 2026. [Paper]
3. Zep: A Temporal Knowledge Graph Architecture for Agent Memory Preston Rasmussen, Pavlo Paliychuk, Travis Beauvais, et al. CoRR 2025. [Paper]

Safety Management

Access Control

1. Collaborative Memory: Multi-User Memory Sharing in LLM Agents with Dynamic Access Control Alireza Rezazadeh, Zichao Li, Ange Lou, et al. CoRR 2025. [Paper]

Privacy Protection

1. Collaborative Memory: Multi-User Memory Sharing in LLM Agents with Dynamic Access Control Alireza Rezazadeh, Zichao Li, Ange Lou, et al. CoRR 2025. [Paper]
2. Privacy-Enhancing Paradigms within Federated Multi-Agent Systems Zitong Shi, Guancheng Wan, Wenke Huang, et al. CoRR 2025. [Paper]

Memory Storage

Textual Memory

Structured Memory

1. ChatDB: Augmenting LLMs with Databases as Their Symbolic Memory Chenxu Hu, Jie Fu, Chenzhuang Du, et al. CoRR 2023. [Paper]
2. MemLLM: Finetuning LLMs to Use An Explicit Read-Write Memory Ali Modarressi, Abdullatif Köksal, Ayyoob Imani, et al. CoRR 2024. [Paper]

Semi-Structured Memory

1. AriGraph: Learning Knowledge Graph World Models with Episodic Memory for LLM Agents Petr Anokhin, Nikita Semenov, Artyom Y. Sorokin, et al. IJCAI 2025. [Paper]
2. Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory Prateek Chhikara, Dev Khant, Saket Aryan, et al. ECAI 2025. [Paper]
3. Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation Zhanghao Hu, Qinglin Zhu, Hanqi Yan, et al. CoRR 2026. [Paper]
4. MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents Dongming Jiang, Yi Li, Guanpeng Li, Bingzhe Li. ACL 2026. [Paper]
5. MemoBrain: Executive Memory as an Agentic Brain for Reasoning. Hongjin Qian, Zhao Cao, Zheng Liu. ACL 2026. [Paper]
6. Zep: A Temporal Knowledge Graph Architecture for Agent Memory Preston Rasmussen, Pavlo Paliychuk, Travis Beauvais, et al. CoRR 2025. [Paper]
7. From Isolated Conversations to Hierarchical Schemas: Dynamic Tree Memory Representation for LLMs Alireza Rezazadeh, Zichao Li, Wei Wei,et al. ICLR 2025. [Paper]

Unstructured Memory

1. LightMem: Lightweight and Efficient Memory-Augmented Generation Jizhan Fang, Xinle Deng, Haoming Xu, et al. CoRR 2025. [Paper]
2. Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory Prateek Chhikara, Dev Khant, Saket Aryan, et al. ECAI 2025. [Paper]
3. Collaborative Memory: Multi-User Memory Sharing in LLM Agents with Dynamic Access Control Alireza Rezazadeh, Zichao Li, Ange Lou, et al. CoRR 2025. [Paper]
4. MIRIX: Multi-Agent Memory System for LLM-Based Agents Yu Wang, Xi Chen. CoRR 2025. [Paper]

Model-Level Memory

Parameter Storage

1. MemOS: A Memory OS for AI System Zhiyu Li, Shichao Song, Chenyang Xi, et al. CoRR 2025. [Paper]

Latent Storage

1. MemOS: A Memory OS for AI System Zhiyu Li, Shichao Song, Chenyang Xi, et al. CoRR 2025. [Paper]

Survey

1. Memory in the Age of AI Agents Yuyang Hu, Shichun Liu, Yanwei Yue, et al. CoRR 2025. [Paper]
2. A survey on large language model based autonomous agents Lei Wang, Chen Ma, Xueyang Feng, et al. Frontiers Comput. Sci. 2024. [Paper]
3. A Survey on the Memory Mechanism of Large Language Model-based Agents Zeyu Zhang, Quanyu Dai, Xiaohe Bo, et al. ACM Trans. Inf. Syst. 2025. [Paper]

Open Problems