This proposal is based on extensive daily use of ChatGPT in a real metal-fabrication business, where designs are ultimately converted into physical manufactured products.
Feature Request: Native AI Vector/CAD Workflow for Manufacturing and Fabrication
I run a professional metal fabrication company in Greece, specializing in custom steel and aluminium constructions, including architectural and ecclesiastical metalwork.
I use ChatGPT extensively as a professional design and product-development tool. From real-world production use, I believe there is a major opportunity for ChatGPT to evolve beyond image generation and become a true AI-assisted CAD and manufacturing design platform.
The current image-generation system is very useful for concepts, but there is a significant limitation for professional fabrication: when we request a small modification to an existing design, the image model may unintentionally modify geometry, proportions, decorative elements, symmetry, or other parts of the construction that were supposed to remain unchanged.
For manufacturing, we need deterministic geometry rather than regenerated pixels.
I believe ChatGPT would become an extremely powerful professional tool if it could introduce a native Vector/CAD layer controlled directly by the AI reasoning model.
Ideally, the workflow could support:
• Vector-based geometry rather than pixel-only image generation.
• Locked geometry, where the user can specify which elements must remain completely unchanged.
• Real dimensions in millimetres.
• Parametric constraints, symmetry, radii, angles and distances.
• Recognition of standard manufacturing materials and profiles, for example flat bar 16×8 mm, rectangular hollow sections, round tube, sheet metal, etc.
• Individual editable components and objects instead of regenerating the entire design.
• Commands such as:
“Move this component 30 mm to the left.”
“Increase this scroll by 15% without modifying anything else.”
“Keep the frame completely locked.”
“Maintain perfect symmetry.”
“Use 16×8 mm flat bar for this element.”
• Identification of real connection and welding points.
• Basic manufacturability checks before approving a design.
• Automatic Bill of Materials / cutting lists.
• Export to formats such as DXF, SVG and eventually DWG/STEP.
• A direct workflow from:
AI concept → Vector geometry → Manufacturing constraints → DXF/CAD → Bill of Materials → Photorealistic render
An especially valuable feature would be the ability to use the same underlying CAD geometry for both the manufacturing drawing and the photorealistic presentation.
This would mean that a fabricator could design a product with ChatGPT, modify it conversationally, obtain an accurate manufacturing file, calculate materials, and then produce a realistic presentation for the customer — without rebuilding the design manually in a separate CAD package.
This would have applications far beyond metal fabrication:
• architectural metalwork
• steel fabrication
• aluminium systems
• furniture manufacturing
• CNC laser and plasma cutting
• sheet-metal fabrication
• woodworking
• industrial design
• mechanical design
• signage
• gates, railings and architectural products
• small and medium manufacturing businesses
For companies like ours, this could dramatically shorten the process from:
idea → design → quotation → customer approval → production.
I strongly believe there is a large professional market for an AI system that does not simply “draw something that looks manufacturable,” but actually understands manufacturing geometry and produces editable, dimensionally accurate designs.
ChatGPT already has the conversational reasoning interface required for this. Connecting that reasoning much more tightly with a deterministic vector/CAD engine could turn it into an exceptionally powerful tool for real-world manufacturing.
This suggestion comes from extensive practical use of ChatGPT in actual metal fabrication and product-development work.
Metallo Xenos – Special Steel & Aluminium Constructions
Greece