Nvidia on Monday introduced NemoClaw, a software stack that deploys its Nemotron models and OpenShell runtime onto the OpenClaw agent platform with a single command.
OpenClaw allows developers to build autonomous AI agents that can complete tasks, delegate work, and access local files. But that level of access also creates risk — compromised agents could delete data or expose sensitive information. NemoClaw is designed to address this by adding sandboxing and policy-based controls for security, privacy, and network access.
Unveiled at Nvidia’s GTC conference, NemoClaw can run models locally or connect to advanced cloud systems through a privacy router. It supports a wide range of hardware, including RTX PCs, workstations, and DGX AI systems.
Jensen Huang said the platform is now “enterprise ready,” describing OpenClaw as an “operating system for personal AI.” OpenClaw creator Peter Steinberger added that the collaboration aims to build powerful but secure AI assistants.
The launch is part of Nvidia’s broader push into autonomous agents, which Huang described as the next major wave in enterprise computing.