Google Cloud 推出 Always-On Memory Agent，用 LLM 持续整合替代 RAG 与向量数据库

大多数 AI 代理会遗忘。它们处理请求后给出答案，然后丢掉上下文。Google Cloud 的生成式 AI 仓库现在提供了一个直接解决这一问题的示例。它就是 Always-On Memory Agent：https://github.com/GoogleCloudPlatform/generative-ai/tree/main/gemini/agents/always-on-memory-agent，一个将记忆视为持续运行进程的参考实现。

从根本上说，该项目是一个轻量级的后台代理，永远不会停止。它作为一个连续运行的进程 24/7 工作，而不是一次性调用。它是使用 Google ADK（代理开发工具包）和 Gemini 3.1 Flash-Lite 构建的。值得注意的是，它不使用向量数据库也不使用嵌入。取而代之的是，LLM 会读取、思考，并将结构化记忆写入 SQLite。模型的选择目标是实现低延迟和低成本的连续后台工作。

在架构上，一个编排器将每个请求分配给三个专业子代理之一。每个子代理拥有自己的工具，用于读取或写入记忆存储。

首先，IngestAgent 处理传入内容。它利用 Gemini 的多模态能力提取摘要、实体、主题和重要性评分。然后，该结构化记录被存入 memories 表中。

接下来，ConsolidateAgent 按计时器运行，默认每 30 分钟执行一次。像睡眠周期一样，它会审查未整合的记忆并寻找它们之间的联系。然后，它将一个综合摘要、一项关键见解和这些联系写入数据库。因此，该代理在空闲时也能构建新理解，无需任何提示。

最后，QueryAgent 回答问题。它读取所有记忆和整合见解，然后综合生成答案。重要的是，它会引用所用记忆的 ID 作为来源。

除了文本，IngestAgent 可接受五类共 27 种文件类型。只需将任何受支持的文件放入 ./inbox 文件夹，即可自动采集。

为了澄清区别，它描述了三种常见的记忆方法。每种方法都解决了部分问题，但仍存在空白。

与 RAG 不同，该代理主动处理记忆，而不仅仅是在检索时。

实际上，这种模式适用于任何需要持久、可演化上下文的工作负载。考虑三个示例。

设计完成后，对于初级工程师来说，设置保持最小化。安装依赖项，设置你的密钥，然后启动进程。

运行后，代理会监控 ./inbox，每 30 分钟进行一次整合，并在端口 8888 提供 HTTP API。因此，你也可以通过 HTTP 向它提供数据。

此外，API 还暴露了 /status、/memories、/consolidate、/delete 和 /clear。可选的 Streamlit 仪表板提供数据输入、查询、浏览和删除控制。CLI 标志可以更改监控文件夹、端口和整合间隔。

需要与我们合作推广你的 GitHub 仓库或 Hugging Face 页面或产品发布或网络研讨会等？请联系：https://forms.gle/wbash1wF6efRj8G58

Michal Sutter 是一名数据科学专业人士，拥有帕多瓦大学数据科学硕士学位。凭借在统计分析、机器学习和数据工程方面的坚实基础，Michal 擅长将复杂数据集转化为可执行的见解。

构建一个代理式活动场地运营商 [完整代码]：https://pxllnk.co/twdn5

谢谢！我们的团队会很快联系你

Most AI agents forget. They process a request, answer it, then drop the context. Google Cloud’s generative-ai repository now ships a sample that tackles this directly. It is the Always-On Memory Agent ：https://github.com/GoogleCloudPlatform/generative-ai/tree/main/gemini/agents/always-on-memory-agent, a reference implementation that treats memory as a running process.

Fundamentally, the project is a lightweight background agent that never stops. It runs 24/7 as a continuous process, not a one-shot call. It is built with Google ADK (Agent Development Kit) and Gemini 3.1 Flash-Lite . Notably, it uses no vector database and no embeddings. Instead, an LLM reads, thinks, and writes structured memory into SQLite . The model choice targets low latency and low cost for continuous background work.

Architecturally, an orchestrator routes every request to one of three specialist sub-agents. Each sub-agent owns its own tools for reading or writing the memory store.

First, the IngestAgent handles incoming content. It uses Gemini’s multimodal capabilities to extract a summary, entities, topics, and an importance score. That structured record then lands in the memories table.

Next, the ConsolidateAgent runs on a timer, every 30 minutes by default. Like sleep cycles, it reviews unconsolidated memories and finds connections between them. Then it writes a synthesized summary, one key insight, and those connections to the database. Consequently, the agent builds new understanding while idle, with no prompt.

Finally, the QueryAgent answers questions. It reads all memories and consolidation insights, then synthesizes a response. Importantly, it cites the memory IDs it used as sources.

Beyond text, the IngestAgent accepts 27 file types across five categories. Simply drop any supported file into the ./inbox folder for automatic pickup.

To clarify the difference, it frames three common memory approaches. Each solves part of the problem, yet leaves a gap.

Unlike RAG, this agent processes memory actively, not only on retrieval.

Practically, the pattern fits any workload needing durable, evolving context. Consider three examples.

With the design clear, setup stays minimal for early-level engineers. Install dependencies, set your key, then start the process.

Once running, the agent watches ./inbox , consolidates every 30 minutes, and serves an HTTP API on port 8888. Therefore, you can also feed it over HTTP.

Additionally, the API exposes /status , /memories , /consolidate , /delete , and /clear . An optional Streamlit dashboard adds ingest, query, browse, and delete controls. CLI flags change the watch folder, port, and consolidation interval.

Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us ：https://forms.gle/wbash1wF6efRj8G58

Michal Sutter is a data science professional with a Master of Science in Data Science from the University of Padova. With a solid foundation in statistical analysis, machine learning, and data engineering, Michal excels at transforming complex datasets into actionable insights.

Build an Agentic Event Venue Operator [Full Codes]：https://pxllnk.co/twdn5

Thanks! Our team will contact you soon