Articraft drops agentic 3D articulated asset generator

Articraft shifts 3D generation from static mesh prompting to an iterative "thinking" workflow that solves the mechanical reasoning gap.

• –Uses an agentic coding loop (GPT-5.5) to write Python code that defines kinematics, joint limits, and PBR materials
• –Outputs URDF files ready for physics engines like NVIDIA Isaac Sim, bypassing manual rigging in Blender
• –Iterative feedback loop with a domain-specific SDK ensures functional correctness (e.g., doors hinge, wheels roll)
• –Solves the data scarcity problem for robotics training by enabling scalable, high-quality synthetic asset generation
• –At ~$1 per asset, it significantly undercuts manual outsourcing for complex simulation environments