A-MEM: Agentic Memory for LLM Agents

2025-02-17

AI Summary

Current LLM agent memory systems predominantly rely on static storage, lacking dynamic organization and task-adaptive capabilities—even graph-based approaches suffer from rigid, predefined schemas. To address this, we propose the first Zettelkasten-inspired, agent-driven memory architecture, enabling autonomous indexing, semantic linking, bidirectional updating, and continual memory evolution. Our method integrates structured note generation, multidimensional attribute annotation, schema-aware relational analysis, and LLM-guided, decision-based memory management to realize adaptive memory network growth. Evaluated across six mainstream foundation LLMs, our approach consistently outperforms state-of-the-art baselines on standard memory-intensive benchmarks. The implementation is fully open-sourced to foster reproducibility and community advancement.

Technology Category

• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Agent/AI Theories and Architectures

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Agentic search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

While large language model (LLM) agents can effectively use external tools for complex real-world tasks, they require memory systems to leverage historical experiences. Current memory systems enable basic storage and retrieval but lack sophisticated memory organization, despite recent attempts to incorporate graph databases. Moreover, these systems' fixed operations and structures limit their adaptability across diverse tasks. To address this limitation, this paper proposes a novel agentic memory system for LLM agents that can dynamically organize memories in an agentic way. Following the basic principles of the Zettelkasten method, we designed our memory system to create interconnected knowledge networks through dynamic indexing and linking. When a new memory is added, we generate a comprehensive note containing multiple structured attributes, including contextual descriptions, keywords, and tags. The system then analyzes historical memories to identify relevant connections, establishing links where meaningful similarities exist. Additionally, this process enables memory evolution - as new memories are integrated, they can trigger updates to the contextual representations and attributes of existing historical memories, allowing the memory network to continuously refine its understanding. Our approach combines the structured organization principles of Zettelkasten with the flexibility of agent-driven decision making, allowing for more adaptive and context-aware memory management. Empirical experiments on six foundation models show superior improvement against existing SOTA baselines. The source code is available at https://github.com/WujiangXu/AgenticMemory.

Problem

• Develops dynamic memory for LLM agents.
• Enhances memory organization using Zettelkasten method.
• Improves adaptability across diverse tasks.

Innovation

• Dynamic memory organization
• Interconnected knowledge networks
• Agent-driven decision making