Nvidia introduces NemoClaw, a new software layer designed to make the OpenClaw platform more secure and accessible.
Nvidia introduces NemoClaw at its own GTC conference. The NemoClaw stack for OpenClaw allows users to install Nemotron models and the OpenShell runtime with a single command. Nvidia provides built-in control mechanisms for privacy and security. Policy-based guardrails regulate the access of autonomous agents.
read also
OpenClaw conquers the world: open source success story or experiment gone wrong?
Users can choose between local models, such as Nemotron, or cloud-based frontier models. A privacy router determines which tasks are executed locally and which go to the cloud. According to Nvidia, this combination ensures that AI agents can learn new skills without compromising security.
Broad support
NemoClaw runs on various Nvidia platforms, including GeForce RTX PCs, RTX PRO workstations, and the DGX Station and DGX Spark supercomputers. The solution is compatible with any coding agent and supports open models. This allows developers to build and customize their own AI assistants according to their specific needs. Nvidia emphasizes that the combination of local and cloud models provides a solid foundation for scalable AI applications.
Never shy of exaggeration, Nvidia’s CEO positions OpenClaw as the ‘operating system’ for AI, comparable to how Mac and Windows are for PCs. With NemoClaw, Nvidia aims to accelerate the adoption of personal AI assistants and, above all, play a central role within the OpenClaw hype.