At a time when artificial intelligence (AI) is transforming many industrial sectors, one question comes up regularly in workshops and design offices:
“Why can’t AI directly generate 2D drawings for the manufacture of mechanical parts?”
On paper, it seems simple: you would just describe a part or provide a sketch, and a well-trained AI would produce a dimensioned, standardised drawing ready to send to the workshop. In reality, things are far more complex.
At SXE-Consulting, we support industrial companies in their digital transformation and in optimising their technical processes. This naturally includes the interface between the physical world (workshop, manufacturing, maintenance) and digital tools (CAD, ERP, MES). Today, let us take stock of the current limitations of AI in generating 2D mechanical drawings.
1. A manufacturing drawing is much more than a sketch
A 2D drawing intended for mechanical manufacturing is not a simple sketch. It must comply with strict standards (ISO, ASME…) and contain precise information on:
-
les geometric dimensions and tolerances,
-
les surface finishes and roughness values,
-
les raw materials and any heat treatments,
-
les welding, threading and machining symbols,
-
les functional and assembly requirements.
This level of detail cannot be improvised. It stems from engineering reasoning that takes into account the mechanical, thermal, environmental and economic constraints of the product.
Today, generative AI (such as ChatGPT or other specialised systems) does not yet spontaneously incorporate all these normative and functional elements, unless it is guided step by step.
2. Lack of technical context and specifications
An industrial drawing cannot be produced generically. The same geometric shape may meet very different requirements depending on:
-
the sector (aerospace, automotive, energy…),
-
the function of the part (transmission, sealing, structure…),
-
the operating environment (temperature, pressure, vibrations…),
-
the available manufacturing processes (milling, moulding, 3D printing…).
AI needs a clear, structured and complete set of specifications to propose a technically viable drawing. Yet these specifications are rarely formulated in a language that an AI can interpret autonomously.
At SXE-Consulting, this is precisely where we help our clients to structure their technical data, so that it becomes usable in digital systems, including with automation or AI tools.
3. AI is not (yet) integrated into industrial CAD tools
Another important limitation lies in the software tools used in industry. CAD (Computer-Aided Design) software such as:
-
SolidWorks
-
CATIA
-
Inventor
-
PTC Creo
-
AutoCAD Mechanical
are complex and highly standardised environments, often coupled with technical data management tools (PDM/PLM).
Today, generative AIs are not natively integrated into these tools. They therefore cannot create a directly usable 2D drawing, that is to say in the form of a DWG, DXF or dimensioned PDF file, in line with industrial practice.
Some solutions are emerging in more recent tools such as Fusion 360 or Onshape, which offer intelligent assistance modules. But we are still a long way from complete and robust automation.
4. Engineering is not just geometry
An AI can imagine a part geometrically. But mechanical engineering also relies on physical principles, calculations and functional reasoning:
-
Will the part withstand the load?
-
Is it compatible with the other elements of the assembly?
-
Can it be machined with the available resources?
-
What is its cost price?
These decisions still fall within the engineer’s profession. AI can assist, suggest and validate certain hypotheses, but it does not (yet) replace the reasoning of an experienced technician or designer.
5. The risks of faulty manufacturing are too high
In an industrial environment, a poorly defined drawing can lead to:
-
un costly scrap,
-
des extended production lead times,
-
une quality non-conformity,
-
or even risks of failure in service.
This is why companies remain cautious and prefer to rely on proven tools and rigorous validation processes.
So what role can AI play today?
Even if AI does not yet generate 2D drawings ready for manufacturing, it can already provide many services:
-
Generation of concept sketches, to help during the design phase,
-
Analysis of existing drawings, to detect inconsistencies or omissions,
-
Translation of business needs into technical specifications,
-
Assistance with functional dimensioning,
-
Searching libraries of standardised components.
And tomorrow, we can imagine:
-
design assistants capable of automatically incorporating tolerances,
-
AI copilots integrated into CAD software,
-
platforms for the automatic configuration of parts based on a customer catalogue.
Conclusion
Artificial intelligence opens up many prospects for the manufacturing industry. But when it comes to the automatic generation of 2D drawings for mechanical manufacturing, we are not there yet. It requires a combination of human skills, integrated digital tools and a normative rigour that AI does not yet fully master.
At SXE-Consulting, we are convinced that the digital transformation of industrial companies will come about through the progressive and intelligent integration of these technologies, in support of existing expertise.
If you would like to discuss the possibilities of automation, MES/ERP/CAD integration, or the structuring of your technical data, contact us. We will help you turn your ideas into concrete, effective projects.
Xavier Schuster
Consultant at SXE Consulting. Industrial consulting firm based in Luxembourg, 25 years of experience in operational excellence.
View profile →