Inspiration
The idea for Cad.ai came from struggling in an engineering class where creating even simple objects in AutoCAD felt unnecessarily difficult. As a beginner, I found CAD software overwhelming powerful, but not very beginner-friendly.
What it does
Cad.ai is an AI-powered assistant that instantly generates AutoCAD designs from your voice commands or uploaded images. By simply stating an idea or providing a sketch, users can go from concept to creation without getting lost in complex menus and features.
How we built it
We built a Next.js frontend connected to:
- An LLM for image classification and idea-to-CAD translation.
- The Web Speech API, with a custom wake word (“Hey Cad”), to capture voice commands.
- A Flask backend, which processes the input and generates AutoLISP files.
We created a listener to detect when an LSP file is generated or updated, which triggers an AppleScript. The script runs the file directly inside AutoCAD, instantly creating the requested object.
Challenges we ran into
Getting AppleScript to properly execute commands in AutoCAD and reliably trigger when new LSP files were created was one of our biggest hurdles.
Accomplishments that we're proud of
We successfully automated the entire workflow so after opening the website, users don’t even need to touch their keyboard. Commands are spoken, interpreted, and executed seamlessly in AutoCAD.
What we learned
We learned how to automate backend processes, integrate Flask with system-level scripts, and combine speech, AI, and CAD into a smooth pipeline.
What's next for CAD.ai
We’d like to train a custom ML model on hundreds of AutoCAD diagrams to improve accuracy and complexity of designs. Ultimately, we envision Cad.ai as a startup that makes CAD more accessible, efficient, and intuitive.
Built With
- applescript
- flask
- gemini
- lisp
- next.js
- openai
- python
Log in or sign up for Devpost to join the conversation.