Paper 2512.13564
Memory in the Age of AI Agents
- Published
- Dec 2025
- Research lab
- Independent
- Citations
- 232
- GitHub
- 2.3K stars
01 In brief
Summary
This survey provides a comprehensive overview of memory systems for foundation model-based AI agents, addressing the fragmentation in the field by proposing a unified taxonomy based on forms, functions, and dynamics.
It defines agent memory, distinguishing it from LLM memory, RAG, and context engineering.
The survey categorizes memory forms into token-level, parametric, and latent memory; functions into factual, experiential, and working memory; and dynamics into formation, evolution, and retrieval.
It compiles benchmarks and open-source frameworks, and discusses future frontiers including automation, reinforcement learning integration, multimodal memory, shared memory, and trustworthiness.
The goal is to establish a conceptual foundation for memory as a first-class primitive in agent design, moving beyond traditional long/short-term distinctions to capture the diversity of modern systems.
The survey is authored by a large team from multiple institutions and is available on arXiv and GitHub, aiming to serve as a reference for researchers and practitioners in the field of agentic intelligence.
It emphasizes the shift from static storage to dynamic, self-evolving memory systems that enable continual learning and adaptation in agents.
The survey also highlights the importance of memory in enabling long-horizon reasoning, continual adaptation, and effective interaction with complex environments, positioning it as a cornerstone of agentic intelligence.
The proposed taxonomy aims to unify emerging concepts and bridge the gap between existing definitions and rapid methodological advances, particularly those from 2025.
The survey also provides a forward-looking perspective on emerging research frontiers, including automation-oriented memory design, deep integration of reinforcement learning, multimodal memory, shared memory for multi-agent…
02 From the paper
Abstract
Memory has emerged, and will continue to remain, a core capability of foundation model-based agents. As research on agent memory rapidly expands and attracts unprecedented attention, the field has also become increasingly fragmented. Existing works that fall under the umbrella of agent memory often differ substantially in their motivations, implementations, and evaluation protocols, while the proliferation of loosely defined memory terminologies has further obscured conceptual clarity. Traditional taxonomies such as long/short-term memory have proven insufficient to capture the diversity of contemporary agent memory systems. This work aims to provide an up-to-date landscape of current agent memory research. We begin by clearly delineating the scope of agent memory and distinguishing it from related concepts such as LLM memory, retrieval augmented generation (RAG), and context engineering. We then examine agent memory through the unified lenses of forms, functions, and dynamics. From the perspective of forms, we identify three dominant realizations of agent memory, namely token-level, parametric, and latent memory. From the perspective of functions, we propose a finer-grained taxonomy that distinguishes factual, experiential, and working memory. From the perspective of dynamics, we analyze how memory is formed, evolved, and retrieved over time. To support practical development, we compile a comprehensive summary of memory benchmarks and open-source frameworks. Beyond consolidation, we articulate a forward-looking perspective on emerging research frontiers, including memory automation, reinforcement learning integration, multimodal memory, multi-agent memory, and trustworthiness issues. We hope this survey serves not only as a reference for existing work, but also as a conceptual foundation for rethinking memory as a first-class primitive in the design of future agentic intelligence.