Is Agent Memory a Database? Rethinking Data Foundations for Long-Term AI Agent Memory

智能体记忆是数据库吗？重新思考长期 AI 智能体记忆的数据基础

Abstract: Long-running AI agents need persistent memory. Memory supports learning across sessions, reduces repeated context injection, and enables auditing of past decisions. Current agent memory systems and database paradigms treat memory as storage. They localize correctness at records, embeddings, or edges. Each supplies only some of the capabilities that long-term memory requires.

摘要： 长时间运行的 AI 智能体需要持久化记忆。记忆支持跨会话学习，减少重复的上下文注入，并能够对过去的决策进行审计。当前的智能体记忆系统和数据库范式将记忆视为存储，并将正确性局限于记录、嵌入或边（edges）层面。每种方法都只能提供长期记忆所需的部分能力。

The result is four recurring failure modes: unregulated growth, missing semantic revision, capacity-driven forgetting, and read-only retrieval. In our vision, long-term agent memory is a new data-management workload. Its correctness is a property of the state trajectory, not of individual records.

其结果导致了四种反复出现的故障模式：不受控的增长、缺失语义修订、容量驱动的遗忘以及只读检索。在我们的构想中，长期智能体记忆是一种新的数据管理工作负载。其正确性是状态轨迹的一种属性，而非单个记录的属性。

We formalize this as Governed Evolving Memory (GEM). GEM replaces record-level database operations with four state-level operators: ingestion, revision, forgetting, and retrieval. Six correctness conditions govern how the state evolves. Three structural observations establish that no record-level system can satisfy these conditions, regardless of the storage model.

我们将此形式化为“受控演进记忆”（Governed Evolving Memory, GEM）。GEM 用四个状态级操作符取代了记录级数据库操作：摄入（ingestion）、修订（revision）、遗忘（forgetting）和检索（retrieval）。六个正确性条件决定了状态如何演进。三个结构性观察结果表明，无论采用何种存储模型，没有任何记录级系统能够满足这些条件。

We realize the abstraction in MemState, a prototype on a property-graph backend. MemState validates feasibility and exposes the gap to a native engine. We outline three research directions that define memory-centric data management as a workload.

我们在 MemState 中实现了这一抽象，这是一个基于属性图后端的原型系统。MemState 验证了其可行性，并揭示了与原生引擎之间的差距。我们概述了三个研究方向，将以记忆为中心的数据管理定义为一种工作负载。