Integrating ChatGPT / GPT-based AI into AutoCAD Looking for Advice or Existing Plugins

[afloareirazvan]

Hello everyone,

I’m exploring ways to integrate ChatGPT or other GPT-based AI tools directly into AutoCAD via an extension, plugin, or add-on.

My goal is to boost efficiency in my CAD workflows using AI/ML, whether that's for automation, design suggestions, code generation (AutoLISP/VBA/.NET), documentation, or intelligent prompts.

I’m open to:

• Existing plugins or APIs that already enable GPT-AI integration in AutoCAD
• Custom development advice (AutoLISP, .NET, Python, etc.) for connecting ChatGPT to AutoCAD
• Machine learning use cases in CAD workflows (design analysis, drafting assistance, etc.)
• General productivity tips using AI in civil/infrastructure design

I’d really appreciate any advice, sample projects, or insights from anyone who’s tried something similar.

Thank you!

[RSomppi]

---

@afloareirazvan wrote:

I’m exploring ways to integrate ChatGPT or other GPT-based AI tools directly into AutoCAD via an extension, plugin, or add-on.

---

What have you found so far?

[pendean]

@afloareirazvan good luck with that: you're about 10-years too early if you are just an end user and not inside/working at "chatgpt" and Autodesk. Apply to work there, it's where this is all happening:

https://www.autodesk.com/careers/overview 

https://openai.com/careers/ 

In the meantime... don't dismiss CLAUDE, GEMINI, COPILOT and other AI offerings not publicly discussed in mass-media that are out there too: brand-names are not always the best option for specialty needs at the end of the day.

[afloareirazvan]

I've actually tried CLAUDE, GEMINI, and COPILOT and am currently focused on integrating AI into my workflow in a customized and specialized way, trained specifically on my processes.

My current approach and goals:

• I'm exploring GPT-based models (like ChatGPT) to work alongside AutoCAD for:
• Code generation (AutoLISP, VBA, .NET)
• Smart drafting assistants
• Automated documentation
• Design logic support (e.g., rule-based validation, layer structuring, etc.)

Why I do not rely on general AI out of the box?

• Generic models are powerful, but often lack contextual awareness of CAD standards, infrastructure norms, or workflow constraints in civil/public projects.
• I see more value in fine-tuned assistants that understand the structure of my drawings, tags, and even naming conventions.

What I'm looking for:

• Real-world examples or tools (plugins/APIs) integrating GPT or LLMs into AutoCAD directly
• Best practices in connecting GPT via AutoLISP, .NET, or Python
• Ideas on embedding local project knowledge into LLM prompts
• Ways to optimize productivity in CAD through AI-driven scripting or suggestions

If anyone here has experimented with:

• Prompt templates for drawing logic
• Using ChatGPT to generate or debug LISP routines
• Embedding metadata from drawings to enrich AI prompts

...I'd love to connect and exchange ideas.

[afloareirazvan]

What I’ve Found So Far

1. Scripting Interfaces for AI Integration
• AutoLISP: Ideal for automating repetitive CAD tasks; easy to generate and test code from AI prompts.
• .NET API (C#/VB.NET): Enables deeper plugin development and custom commands inside AutoCAD.
• Python (via pyautocad or COM interfaces): Allows external scripting and communication between AutoCAD and APIs like OpenAI.
2. Tools and Libraries Used
• OpenAI API: For generating code (AutoLISP, VBA, .NET), documentation, and intelligent prompts.
• pyautocad: A Python library to control AutoCAD through COM, useful for connecting CAD logic to AI-generated instructions.
• ChatGPT: Used manually for drafting assistance, generating macros, or validating code logic.
3. Experimental Methods
• Using GPT to generate AutoLISP scripts from natural language prompts.
• Python scripts that:
• Extract drawing data (e.g., block names, attributes, layer structures).
• Send the data to GPT for suggestions or logic checks.
• Return draft scripts or tag instructions for implementation.
• Prompt chaining for common CAD workflows (e.g., automating drawing setup, annotation generation, or dimensioning logic).

Next Steps I’m Exploring

• Building a lightweight GPT-based assistant that understands:
• My project templates
• CAD standards and layer naming conventions
• Drawing metadata and workflows
• Training or fine-tuning models on internal project data (to avoid relying on general-purpose responses).