01 What happened
NVIDIA describes how engineers direct frontier AI models such as Astra and Claude Fable 5 through natural-language prompts to build simulation applications and OpenUSD digital twins. The post gives no pricing or general availability dates.
02 Key details
- NVIDIA reports that Ashley Reid directed Astra and Claude Fable 5 agents over about three days. The agents created two digital twins from scratch and improved two existing ones, comparing camera and raw LiDAR outputs with recorded data.
- Frank DeLise directed Astra to connect Omniverse libraries for physics, scene updates, rendering and the user interface while building a humanoid warehouse simulator.
- In one NVIDIA-reported experiment, a simulated robot cleared a single hurdle in 64 of 100 simulation trials. The result is company-reported and not independently verified.
- Jens Jebens reported successful removal of a suspension component in simulation after an AI agent designed a wrench the robot could use to reach the suspension's bolts.
03 Why it matters
For robotics, simulation and digital twin teams, this shows how natural-language direction of AI agents could shorten the building and testing of environments. It is a vendor account of internal projects, not proof of production results or broader availability.
04 Who it matters to
Robotics system developers, simulation developers and digital twin developers.
Original source