Details
- NVIDIA has expanded its Agent Toolkit to include NVIDIA Omniverse libraries, enabling AI agents to add physical AI and simulation-ready capabilities to existing 3D and CAD applications.
- Key components include ovrtx for RTX sensor simulation, ovphysx for GPU-accelerated physics, and CAD-to-SimReady skills for converting CAD data into OpenUSD-based SimReady assets, all openly available on GitHub.
- Software makers such as SideFX and PTC, along with startups ForgeCAD, Lightwheel, Moonlake AI and Palatial, are integrating these libraries to bring agentic AI into tools like Houdini, Onshape and custom 3D content pipelines.
- A "SimReady" Blender blueprint demo, combining Omniverse libraries with NVIDIA NemoClaw, showcases how developers can embed agent-ready sensor simulation, physics and validation directly into existing 3D applications.
- The workflows can run on local and cloud systems, from upcoming RTX Spark PCs by major OEMs to GB300-powered DGX Station systems, targeting robotics, industrial digital twins and other physical AI simulation use cases.
Impact
By turning Omniverse’s sensor, physics and asset-validation capabilities into agent-callable tools, NVIDIA is lowering the barrier for developers to build simulation-first workflows for robots, factories and autonomous systems. This move strengthens NVIDIA’s position at the center of the physical AI stack and is likely to accelerate adoption of agentic tooling across industrial design, robotics simulation and synthetic data pipelines over the next 12–24 months.
