Nvidia has introduced NemoClaw, an open-source framework designed to help enterprises build autonomous AI agents that can handle complex, multi-step engineering tasks without constant human supervision. The toolkit includes specialized blueprints, AI models, a secure runtime called OpenShell, and access to domain-specific software libraries that allow agents to perform engineering work alongside human teams.

What Makes NemoClaw Different From Other AI Agent Tools?

NemoClaw addresses a critical gap in enterprise AI: most AI tools today are designed for quick, single-turn interactions, like chatbots answering questions. But engineering workflows are different. A chip designer might need an AI agent to spend days orchestrating simulations, debugging results, and iterating on designs. NemoClaw is built specifically for these long-running, complex tasks.

The framework includes several key components working together. At its core is OpenShell, an open-source runtime that governs how agents access files, networks, and tools while enforcing security policies at every layer. This matters because enterprises need to control what data their AI agents can touch and where that data can go. Nvidia is also releasing Nemotron 3 Ultra, a 550-billion-parameter mixture-of-experts model optimized for long-running agent tasks in coding, research, and enterprise workflows. The company claims this model delivers up to five times faster inference speed and up to 30% lower costs compared to other frontier models in the same class.

Which Companies Are Already Building AI Engineers With NemoClaw?

The adoption is moving fast across industrial software leaders. Cadence is building an autonomous register-transfer level (RTL) engineer, a specialized AI that handles digital circuit design verification. The workflow compresses RTL verification from weeks down to hours, a dramatic acceleration for a task that typically requires multiple human engineers. Dassault Systèmes is integrating NemoClaw into its 3DEXPERIENCE platform for long-running agents across design, simulation, and manufacturing operations. Siemens is embedding it into Fuse EDA AI Agent, designed to orchestrate multi-tool workflows across semiconductor and circuit board design. Synopsys is collaborating with Nvidia to apply agents to end-to-end engineering workflows, with demos showing autonomous AI engineers optimizing GPU cooling designs.

Beyond traditional engineering software, startups are building specialized AI agents on NemoClaw. Flexcompute is using OpenShell for its Tidy3D and PhotonForge agents, which combine optical, electrical, and thermal simulation to explore thousands of design variants overnight. Luminary is building an agent that reduces the time and complexity of training AI physics models by autonomously orchestrating data generation and machine learning model selection. Neural Concept deployed an agent for electric motor design that chains electromagnetic, structural, and noise simulations in a single pipeline. nTop is using NemoClaw to compress days of geometry iteration into hours for aircraft design. PhysicsX partnered with Microsoft to build an electronics thermal simulation agent that automates the full thermal simulation lifecycle for consumer devices like Microsoft Surface laptops. P-1 AI is building Archie, an AI mechanical and electrical engineer already working with data center cooling and critical power systems. SimScale is adopting NemoClaw to build autonomous simulation agents for hundreds of engineering use cases, including noise and vibration analysis.

How to Deploy NemoClaw Across Your Organization

• Choose Your Deployment Environment: NemoClaw can be deployed from Nvidia DGX Spark personal AI supercomputers, enterprise data centers, or cloud service providers, giving organizations flexibility in where and how they run their agents.
• Select Your Orchestration Framework: NemoClaw works with multiple agent orchestration frameworks including OpenClaw and Hermes Agent, allowing enterprises to integrate it into existing AI infrastructure and workflows.
• Configure Security and Privacy Controls: Use OpenShell to set policy-based controls governing how agents access files, tools, and memory. The runtime can route queries to local models based on user privacy settings and mask personal information before requests go to cloud systems.
• Leverage Domain-Specific Libraries: Access Nvidia's CUDA-X libraries as agent skills, including cuDF for data processing, cuOpt for routing and scheduling, NeMo for optimization, and PhysicsNeMo for scientific simulation.

Nvidia is also working with major infrastructure partners to expand support. Canonical plans to integrate OpenShell with Ubuntu through supported snaps and rocks, while Red Hat is integrating it into Red Hat AI and contributing to the upstream open-source project. SAP has embedded OpenShell into the Joule Studio runtime in its business AI platform, and ServiceNow is using it to secure Project Arc, its autonomous desktop agent, with policy-based management controls.

Why Security and Privacy Matter for Long-Running Agents

As AI agents gain more autonomy and access to sensitive systems, security becomes critical. OpenShell addresses this by enforcing identity, containment, policy, and end-to-end security controls. Nvidia is also working with Microsoft on a native Windows environment for personal agents using new Windows security primitives alongside OpenShell. This partnership ensures that agents running on consumer devices can operate securely without exposing user data.

"The world's software leaders are bringing AI agents into the systems where work gets done, showing how AI coworkers help employees think faster and execute complex tasks to solve bigger problems. NVIDIA NemoClaw provides enterprise software developers with the open building blocks to create more secure, long-running AI coworkers that amplify human expertise as they reshape how work gets done," said Jensen Huang, founder and chief executive officer of Nvidia.

Jensen Huang, Founder and Chief Executive Officer, Nvidia

NemoClaw is available now, while OpenShell is in early preview. The framework represents a shift in how enterprises think about AI, moving beyond chatbots and single-turn interactions toward AI coworkers that can handle weeks-long engineering projects with minimal human intervention. For industries like semiconductors, aerospace, and manufacturing, where simulation and design verification consume enormous amounts of time and resources, this capability could fundamentally change how work gets done.