Adam, a startup focused on AI‑driven mechanical design, released CADAM, an open‑source text‑to‑CAD web app. Built with React 19 and Tan Stack Start, the platform lets users describe a part in plain English or upload an image, then instantly generates parametric 3‑D models. The output appears in a Three.js preview and can be exported as .STL, .SCAD, or .DXF.

Behind the scenes, Supabase handles authentication, storage, and database needs, while OpenSCAD code is compiled to WebAssembly and run in a web worker, keeping the UI snappy. The AI engine is agnostic; it can call Anthropic Claude, Google Gemini, or OpenAI via OpenRouter, with Gemini 3.1 Pro topping benchmarks for speed and accuracy.

Users can tweak dimensions through interactive sliders that update the SCAD source with deterministic regex, bypassing the LLM. The tool supports BOSL, BOSL2, and MCAD libraries and lets designers embed custom fonts like Geist. Planned upgrades aim to add build123d and CadQuery support, moving from CSG primitives to constraint‑driven models, and enhancing spatial context with viewport‑guided LLM prompts.

CADAM is open source under GPL‑3.0, inviting contributions from the community. Developers can clone the GitHub repo, run Supabase locally, and start the dev server with npm scripts. The project demonstrates how AI can streamline mechanical design workflows, lowering entry barriers for hobbyists and rapid prototyping teams.