Adam Introduces Open-Source AI-Powered CAD Generation

Adam, a startup incubated through Y Combinator W25, has launched CADAM, an open-source platform that uses artificial intelligence to generate parametric 3D models from natural language or image inputs. The project aims to transform mechanical design by treating it as code—similar to how software development operates today.

The platform offers a user-friendly interface where users can input text prompts or upload images to create 3D designs. CADAM then generates OpenSCAD code, which forms the foundation of the model and includes interactive sliders that allow for real-time dimension adjustments without needing to re-prompt the AI. Designs can be exported in multiple formats including STL, SCAD, OBJ, GLB/GLTF, FBX, and DXF.

Technical Highlights:

• Architecture: React app (TanStack Start) with a Supabase backend for authentication, database, and file storage
• Underlying AI: Single agent endpoint that switches between parametric (OpenSCAD generation) and mesh modes based on input type
• Model Support: Anthropic Claude, Google Gemini, OpenAI models via OpenRouter with adaptive thinking enabled
• Rendering: Uses WebAssembly to compile OpenSCAD in the browser with a responsive React Three Fiber interface
• Libraries Supported: BOSL, BOSL2, MCAD plus custom font support (Geist)

The team emphasizes that CADAM is designed to be extensible, and contributions from the open-source community are encouraged. They plan future improvements including support for additional modeling frameworks like build123d and CadQuery, as well as enhanced spatial context awareness for more complex designs.

See a demo of CADAM in action here: https://www.youtube.com/watch?v=iESOr7EGWqk