amagine-ai/Amagine3D

What it solves

Amagine3D automates the creation of editable 3D hardware designs, specifically focusing on intelligent enclosures and assembly structures. It eliminates the manual effort of translating product descriptions, reference images, and physical dimensions into complex parametric CAD models, while ensuring the resulting designs are manufacturable and editable.

How it works

The system uses a "3D-native Agent" that operates in an autonomous inner loop to generate and refine designs.

1. Requirement Analysis: The Agent converts natural-language descriptions and images into a design brief.
2. Code Generation: It generates parametric CAD source code using the build123d Python library.
3. Verification: The code is run in a browser geometry runtime to build the model. The Agent then performs automated checks for part connectivity, interference, motion paths, and real dimensions.
4. Iterative Refinement: If checks fail, the Agent uses the actual geometric measurements to locate errors and modify the source code until the requirements are met.
5. Commit: Once validated, the design is saved as a new version with its source code and exportable files (STEP, STL, 3MF).

Who it’s for

Hardware designers, engineers, and enthusiasts who need to quickly prototype enclosures and mechanical structures from conceptual requirements without starting from scratch in CAD software.

Highlights

• Editable Source: Produces full Python/build123d source code, allowing users to adjust key dimensions in a workbench without re-running the AI model.
• Automated QA: Checks for collisions, operating clearances, and printing tolerances during the generation process.
• Multi-format Export: Supports exporting to STEP and STL for single-color designs, and color-aware 3MF files for multi-color designs.
• 3D-native Architecture: Uses a state-based approach where the Agent reads actual geometry results rather than relying on textual descriptions of the model.