Jakarta, INTI - Nvidia, a U.S.-based semiconductor and artificial intelligence company, introduced a new technology called NemoClaw during its annual NVIDIA GTC (GPU Technology Conference) 2026, held in San Jose, California, from March 16–19.
NemoClaw is a software stack designed for the OpenClaw platform, an open-source system that allows users to create their own autonomous AI agents. Within this ecosystem, each AI agent is referred to as a “claw,” making OpenClaw essentially a collection of customizable AI agents that users can build and operate.
With NemoClaw, users can develop AI agents that function like automated digital assistants, capable of tasks such as writing code, building applications, and independently handling various computer-based activities.
Nvidia CEO Jensen Huang explained that the launch of NemoClaw is closely tied to the rapid growth of OpenClaw, which he described as one of the fastest-growing open-source projects today. He compared OpenClaw to a computer operating system, but specifically for AI assistants, stating, “Mac and Windows are operating systems for personal computers. OpenClaw can be seen as an operating system for personal AI.”
OpenClaw creator Peter Steinberger welcomed the introduction of NemoClaw, emphasizing that the technology aims to bring AI closer to users. He noted that, together with Nvidia and the broader ecosystem, it enables the development of powerful and secure AI agents, complete with robust guardrails that allow anyone to create reliable AI assistants.
Enabling Secure, Local, and Always On AI Agents
NemoClaw is designed to simplify the creation of AI agents by allowing users to install the entire system with a single command. Once installed, it automatically deploys essential components within the OpenClaw platform, including Nvidia’s Nemotron AI models, OpenShell as a secure AI runtime environment, and additional security and privacy layers. This integrated setup enables AI agents to operate within a sandboxed environment, enhancing data protection.
The platform also includes policy-based security controls, ensuring that AI agents can only access specific data or networks according to user-defined rules.
One of NemoClaw’s key advantages is its ability to run AI models directly on users’ local devices, reducing reliance on cloud infrastructure. However, compatible hardware is required, such as PCs and laptops equipped with Nvidia GeForce RTX GPUs, professional workstations with Nvidia RTX PRO GPUs, or dedicated AI systems like Nvidia DGX Station and DGX Spark. Running AI locally provides a higher level of data privacy compared to fully cloud-based processing.
For more advanced capabilities, AI agents can still connect to cloud-based models through a system called the Privacy Router. In addition to local deployment, NemoClaw is designed to support always-on AI agents, allowing them to continuously perform tasks such as coding, managing data and documents, building simple applications, and assisting with administrative work.
Conclusion
Nvidia’s NemoClaw represents a major step toward making AI agents more accessible, secure, and practical for everyday use. By combining open-source flexibility with powerful local and cloud capabilities, the platform opens new possibilities for individuals and businesses to build and deploy intelligent, autonomous systems.
Read more: Indonesia–Guangxi Collaboration Strengthened, Focusing on Vocational Education and AI Development