TRINIX generates manufacturable, precision 3D CAD models from text or image inputs and allows users to revise selected dimensions and structures through conversation. Unlike mesh data that primarily represents appearance, the generated CAD defines design parameters such as length, curvature, radius, and tolerance. Rather than generating a visual model alone, the technology connects manufacturable CAD, conversational editing, and SimReady data in one workflow—a defining advantage in removing the Physical AI training-data bottleneck.

The data can also include component hierarchies, articulated structures, hinges, mass, friction, and other physical information required for robot simulation. By changing parameters such as size, structure, or opening angle, users can quickly create large sets of variations for synthetic reinforcement-learning data.

Physical AI must learn motion and interaction in addition to visual perception. NdotLight is therefore developing TRINIX as conversational design infrastructure that connects early-stage design with SimReady data generation and robot training, rather than as a standalone 3D generation tool.

TRINIX is currently offered to enterprise customers, including on-premises deployments for organizations with strict security requirements. NdotLight will continue expanding access to high-quality CAD data while improving generation speed and precision.