Open-source AI tool generates parametric CAD models in seconds

CADAM is an open-source AI CAD tool that generates fully parametric CAD models from plain-language descriptions, directly in the browser with no installation required. The project, backed by Y Combinator, supports export to STL, SCAD, and DXF formats and is built for the 3D printing and CAD community.

---

CADAM is an open-source AI-powered CAD tool that allows users to generate fully parametric CAD models from plain-language text descriptions, directly within a web browser. The project requires no installation and is accessible via adam.new/cadam, making it straightforward for users to get started without any local setup.

The tool is capable of producing models ranging from full multi-part machines down to individual parametric parts. All models are generated as fully parametric OpenSCAD, meaning they can be further edited and refined after generation. Completed models can be exported in several formats, including STL, SCAD, and DXF.

CADAM is a Y Combinator W25 company and has been built with the 3D printing and CAD community in mind. The project is open source and licensed under the GNU General Public License v3.0 (GPLv3).

How It Works

CADAM takes a single plain-language prompt as input and produces a parametric CAD model from it. A set of benchmark examples is included in the repository, each showing the original prompt alongside the resulting model, giving prospective users a clear sense of the tool's capabilities. Orbiting previews of benchmark models are rendered using a script included in the repository.

On the technical side, CADAM uses public URLs for provider callbacks and local signed storage URLs. For local development, the project uses ngrok to create tunnels pointing to the development server, and environment variables are used to configure the setup. If a provider needs to fetch local storage signed URLs, a second tunnel can be configured accordingly.

The project is built on TanStack Start and integrates with Supabase Storage for local signed URL handling. Developers wishing to run the project locally are guided through the configuration process via the repository's documentation, including a contributing guide and code of conduct.

Getting Started and Contributing

CADAM is available to use directly in the browser at adam.new/cadam, with no installation needed for end users. Those wishing to self-host or contribute to the project can access the full source code on GitHub, where the repository includes setup instructions, a contributing guide, and a code of conduct.

The project welcomes community contributions via pull requests and issues on the repository. The team has acknowledged the open-source components and tools that made the project possible, and encourages users who find it useful to star the repository.

Story based on discussion on Hacker News.