---
格式版本: 2
标题: "Building a Memory-Driven Agent with NVIDIA NemoClaw | NVIDIA Technical Blog"
原文链接: "https://developer.nvidia.com/blog/building-a-memory-driven-agent-with-nvidia-nemoclaw/"
发布日期: "2026-09-04"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:configured_publication_date_rule"
发布时间证据: "nvidia-developer-blog-published-meta text:scrape: Published: 2026-09-04"
发布时间校准原因: "信源发布日期识别规则直接确认发布时间"
发布时间校准置信度: "high"
发布时间候选数量: 1
发布时间严格候选数量: 1
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-09-06T10:19:50+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-09-06T10:18:04+08:00"
入库时间: "2026-09-06T02:25:32.759Z"
来源平台: "固定入口"
搜索渠道: "fixed_url"
搜索词: "https://developer.nvidia.com/blog/"
匹配关键词:
  - "AI"
  - "performance"
相关厂家:
  - "NVIDIA"
相关专家:
  []
内容类型: "网页"
抓取工具: "CDP Render"
清洗工具: "CDP Text + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 35
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "正文主线是使用NVIDIA NemoClaw构建记忆驱动agent，属于软件代理记忆设计，不涉及超节点、AI Rack或机架级系统；来源为NVIDIA官方博客权威性高，但主题完全不属于机架级AI基础设施；无新增超节点相关标准、产品或部署信号；命中应用/模型效率及教程类强否决项。总分35分，非优质。"
AI质检模型: "zj-deepseek-v4-flash"
AI质检时间: "2026-09-06T10:28:23+08:00"
AI主题相关性: 0
AI来源权威性: 15
AI新颖性: 5
AI技术细节: 5
AI商业部署信号: 0
AI完整性: 10
AI评分提示词版本: "v17-精简生产版"
AI评分提示词SHA256: "48fb9777f386026761b4873eaff30807694fb11e9b352d7c69bf2dfde750cc7d"
AI评分知识库版本: "knowledge_base_v1-20260819+runtime.96"
AI评分知识库SHA256: "0cae74ad2e30b227036c5952745542146844722958bc01868bbd0799ab59eeec"
AI评分知识库检索词: "[\"NVIDIA\",\"https://developer.nvidia.com/blog/\",\"NPU\",\"NVIDIA/nemoclaw-community\",\"SQLite\",\"RAG\",\"MCP\",\"github.com/NVIDIA/nemoclaw-community/issues/122\"]"
AI评分知识库命中: "[{\"id\":\"july-correct-0088\",\"title\":\"StrataCL: Fabric-Native Communication Library for Production Supernodes - 智源社区论文\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"NVIDIA\",\"NPU\"],\"rank\":-7.611542549966371},{\"id\":\"runtime-676f90ee5f88ebab3b504077\",\"title\":\"NVIDIA NVLink Fusion Brings NVHBM to Next-Generation AI Infrastructure\",\"sourceType\":\"ai_excellent_article\",\"time\":\"\",\"matchedTerms\":[\"NVIDIA\",\"https://developer.nvidia.com/blog/\"],\"rank\":-6.548478964443584},{\"id\":\"runtime-5d07dc7b6b3ddbca50e6572d\",\"title\":\"NVIDIA BlueField-4 Powers New Scale-In Network Infrastructure for Agentic AI Factories\",\"sourceType\":\"ai_excellent_article\",\"time\":\"\",\"matchedTerms\":[\"NVIDIA\",\"https://developer.nvidia.com/blog/\",\"RAG\"],\"rank\":-6.520669380005036},{\"id\":\"runtime-f42510f490b4e8271c10e963\",\"title\":\"NVIDIA Vera Rubin and Blackwell Set a New Standard for Agentic AI Performance per Watt\",\"sourceType\":\"ai_excellent_article\",\"time\":\"\",\"matchedTerms\":[\"NVIDIA\",\"https://developer.nvidia.com/blog/\",\"NPU\",\"RAG\"],\"rank\":-6.4530644783571685},{\"id\":\"runtime-0fc9a77588762731c0782840\",\"title\":\"Solving Agentic AI Fleet Challenges with NVIDIA Vera CPU\",\"sourceType\":\"ai_excellent_article\",\"time\":\"\",\"matchedTerms\":[\"NVIDIA\",\"https://developer.nvidia.com/blog/\",\"NPU\",\"RAG\"],\"rank\":-6.432921797387001}]"
采集批次: "2026年9月6日10点16分11秒"
采集批次ID: "20260906-101611-fd9dfd98"
去重键: "https://developer.nvidia.com/blog/building-a-memory-driven-agent-with-nvidia-nemoclaw"
---

Enterprise work spans messages, decisions, projects, and obligations that change over time. An [AI agent](https://www.nvidia.com/en-us/ai/) that starts without this context must reconstruct it before contributing.

To provide agents with this necessary context, our team used [NVIDIA NemoClaw](https://github.com/NVIDIA/NemoClaw) to build a [memory-driven Chief of Staff](https://github.com/NVIDIA/nemoclaw-community/tree/main/examples/recipes/nvidia/memory-driven-chief-of-staff). It maintains a human-readable knowledge layer called the *self model*: an agent memory of relevant people, projects, priorities, and working patterns. Scheduled jobs periodically review new activity, track obligations, and incorporate user decisions over time. Our experience shows that useful agent memory requires structure, selective retrieval, and governance—not just storage.

This post demonstrates how a memory-driven agent built with NVIDIA NemoClaw can improve productivity in real-world enterprise workflows. It also shares five design lessons you can apply to your own agents:

- Maintain context across daily work to improve task quality
- Separate evidence, knowledge, and actions to help agents make better judgments
- Design memory-driven agents can prioritize user intent over short-term urgency
- Allow users to correct agent decisions to build trust
- Enforce the boundary of security and authorization with [NVIDIA OpenShell](https://github.com/NVIDIA/openshell)

## Maintain context across daily work

Conversation history provides short-term continuity, but it mixes current priorities with past decisions and temporary requests. Retrieval can find relevant source material, but the agent must still connect information across time.

Consider a project status question. The answer might depend on an earlier decision, a correction in a later message, an unresolved obligation, and the knowledge that two different names refer to the same project.

To solve these problems, you can use the self model that maintains these relationships in structured Markdown pages. It organizes information about people, projects, priorities, goals, concepts, and recurring work patterns. Its schema defines indexing, cross-references, provenance, and growth limits.

The self model stores a derived interpretation rather than replacing source evidence. Keeping the two separate helps you, the developer, determine whether an incorrect answer came from the evidence, memory maintenance, retrieval, or the model’s final decision.

## Separate evidence, knowledge, and action

You could use the three layers of tuning capabilities from the memory-driven Chief of Staff:

Evidence → Knowledge → Governed execution

Evidence supports updates to the self model. For each task, the agent retrieves a bounded set of relevant context. The agent then uses that context within the NVIDIA NemoClaw example. The architecture is shown in Figure 1.

![A three-step workflow diagram showing how an NVIDIA NemoClaw system uses daily work inputs to create and improve a self model over time. It starts with Daily Work inbox/calendar/tasks feeding into a central self model, then produces results and evidence, which loops back for ongoing learning. The output is then used to generate NemoClaw plans, reasons, and acts within a defined policy/safety and secure action boundary.](https://developer-blogs.nvidia.com/wp-content/uploads/2026/09/nemoclaw-inputs-self-model.webp)

Figure 1. NVIDIA NemoClaw uses persistent context for governed agent actions

The example stores two kinds of information:

- **Knowledge**: People, projects, priorities, and working patterns
- **Judgment**: Whether an item needs attention, where it ranks, and whether the user ignored it

A Markdown agent memory stores knowledge. A SQLite ledger stores obligations, rankings, corrections, and audit events. This design preserves the agent’s judgments without writing them into source messages as read flags, labels, or folders.

A memory page might say that a collaborator prefers Slack. The agent can use that context to recommend Slack, but sending a message still depends on credentials, tool permissions, runtime policy, and user approval.

Context can inform an action, but it cannot authorize one.

## Prioritize user intent over short-term urgency

Incoming requests often describe themselves as urgent, but urgency does not necessarily reflect the user’s priorities. An intent gate therefore reserves the highest tier for obligations connected to the user’s stated priorities.

In the public recipe provided, an urgent expense-policy attestation remains visible but ranks below a quieter request tied to a stated priority. Your NVIDIA NemoClaw could interpret that relationship, while deterministic code enforces tier size, overflow behavior, and ranking order.

## Allow users to correct the agent

Persistent memory can preserve an incorrect judgment as easily as a correct one. Using the recipe, you could move an obligation to another tier or ignore it, and later agent runs preserve that decision. Each change is recorded once in an append-only audit trail.

Repeated correction patterns can update a small, readable preference policy. Users can inspect, edit, or delete that policy instead of leaving the preference hidden in model state.

The feedback loop remains visible:

Agent judgment → User correction → Audit event → Preference update

## Add memory to improve agent task performance

Adding the memory-driven Chief of Staff to NemoClaw produced measurable improvements across several agent tasks, as shown in Table 1. The [Agent Memory Benchmark](https://github.com/NVIDIA/nemoclaw-community/tree/main/examples/tools/agent-memory-benchmark) and evaluation examples are included in the example repo. The example repo compares an agentic [retrieval-augmented generation (RAG)](https://www.nvidia.com/en-us/glossary/retrieval-augmented-generation/) baseline that performs multi-round retrieval with the self model.

| **Metrics** | **Question counts** | **Agentic RAG baseline** | **Self model** | **Difference** |
| --- | --- | --- | --- | --- |
| Overall accuracy | 186 | 82.8% | 90.9% | **+8.1 pp** |
| Hard questions | 31 | 67.7% | 87.1% | **+19.4 pp** |
| Tracking facts that changed over time | 5 | 60.0% | 100.0% | **+40.0 pp** |
| Point-in-time reasoning | 6 | 33.3% | 66.7% | **+33.3 pp** |
| Entity disambiguation | 15 | 66.7% | 86.7% | **+20.0 pp** |
| Multisource synthesis | 73 | 87.7% | 94.5% | **+6.8 pp** |
| Answered faithfully based on Corpus | 13 | 100.0% | 92.3% | **\-7.7 pp** |
| Single-hop lookup | 30 | 86.7% | 83.3% | **\-3.3 pp** |
| Citation coverage | 186 | 92.5% | 97.8% | **+5.4 pp** |

*Table 1. Evaluation metrics for the agentic RAG baseline and the self model in the example repo. Both configurations use NVIDIA Nemotron 3 Ultra*

## Enforce the boundary at runtime

This separation is enforced with NVIDIA NemoClaw and the NVIDIA OpenShell secure runtime for autonomous agents. NemoClaw integrates the example with NVIDIA OpenShell and manages its lifecycle, while NVIDIA OpenShell runs the agent in a sandbox and provides governance and policy enforcement for file system, process, and network access. For managed inference and MCP connections, credentials remain outside the sandbox.

This is important because memory and retrieved content are inputs to the model, not trusted security policy. If the agent misinterprets that context—or follows malicious instructions—it still operates within operator-defined runtime boundaries. They limit the agent’s access and the potential impact of a failure.

## Get started building agent memory

To adapt the memory design featured in this post for your own NemoClaw example, review the open source [Memory-Driven Chief of Staff recipe](https://github.com/NVIDIA/nemoclaw-community/tree/main/examples/recipes/nvidia/memory-driven-chief-of-staff) and its [design proposal](https://github.com/NVIDIA/nemoclaw-community/issues/122) in the NVIDIA/nemoclaw-community GitHub repo.

The recipe packages the example as a deployable Hermes profile for NemoClaw:

- A structured memory schema
- A durable obligation ledger
- Bounded ranking logic
- User correction and audit paths
- Scheduled memory maintenance
- Synthetic messages and memory pages
- An offline walkthrough
- A suite of unit tests

The sample people, organizations, projects, and messages are invented. Recorded model decisions stand in for inference in the offline walkthrough. The code then applies the ranking, correction, persistence, and validation behavior.

The current recipe focuses on the memory foundation. It does not send messages or modify source systems. This scope allows you to examine the design without connecting a workplace account. Live connectors require separate handling for credentials, privacy, retention, and deletion.

Learn more about [NVIDIA NemoClaw](https://docs.nvidia.com/nemoclaw/latest/) and [NVIDIA OpenShell](https://build.nvidia.com/openshell).