With a few references, the right words, and a well-structured prompt, we can generate remarkably compelling product visuals within minutes. AI-powered tools are bringing significant speed to the design process—allowing us to explore form alternatives, investigate different design languages, test material and color combinations, and quickly evaluate the potential of an idea.
And the results are often highly convincing.
Sharp surfaces, seamless transitions, compelling materials, refined details… The product can look as if it has already come off the production line. But this is exactly where an important distinction emerges:
A product that looks manufacturable is not necessarily manufacturable.
What Is Visually Possible Is Not Always Physically Possible
A surface that appears to be only 1 mm thick in a concept visual may look remarkably strong and refined. In a real product, however, we need to determine what material that surface will be made from, how it will behave under load, how it will be supported, and which manufacturing process can actually produce it.
Two parts may appear to meet perfectly in a visual. In reality, manufacturing tolerances, assembly requirements, joining methods, and production constraints can all affect that seemingly perfect transition. A control element may be positioned on an exceptionally clean surface. But behind it, there still needs to be sufficient space for the PCB, wiring, connectors, fasteners, and service access.
The outcome of industrial design is therefore not simply a form.
Form; It is the physical result of decisions between user, engineering, production and brand.
‘ Products Are Designed Around Constraints ‘
In design, the word “constraint” is often perceived as the opposite of creativity.
Yet many successful products emerge precisely from the intelligent management of constraints. Not every parameter shaping a physical product is an obstacle. Some establish boundaries, some determine priorities, and others create entirely new design opportunities. When developing a product, there are many criteria that work at the same time behind the form:
An injection-molded plastic part requires appropriate draft angles. Sheet metal comes with specific bending radii. CNC machining is constrained by tool access. Extrusion requires geometry that can continue consistently along a section. Electronic components occupy defined volumes and have specific connection orientations. In a sealed product, even surface continuity may be more than an aesthetic decision.
None of these are engineering details to be added after the design is complete.
Inputs that shape the form itself.
In a well-integrated design process, manufacturing methods, ergonomics, component placement, and assembly architecture are embedded in design decisions from the earliest stages of form development.
Manufacturability, therefore, should not be treated as a separate phase in which engineering attempts to make an already-defined form feasible. Material selection, manufacturing processes, part architecture, assembly, tolerances, and technical requirements should all contribute to shaping the concept as it develops.
Otherwise, what we have is not a production-ready product design, but a product idea whose path to production has yet to be resolved.
Where Does AI Fit Into the Manufacturability Equation?
For now, AI is not a replacement for manufacturability expertise. It is a tool that significantly expands our capacity for ideation, exploration, and visualization within the design process. AI-powered visualization tools allow us to investigate far more form directions for the same design problem, compare different design languages rapidly, and make ideas visible in a fraction of the time previously required for sketching and visualization.
That is a significant advantage. But what is accelerating most today is the generation, diversification, and visualization of ideas. The design and engineering decisions required to transform an idea into a physical product still need to be made.
An AI image may offer us a new surface relationship. The designer should ask the following questions: What material and production method can this surface create in the actual product?
AI may suggest an unexpected combination detail: How are these parts actually produced and how are they mounted to each other?
It can form a fairly thin and light looking body: How does this body provide the necessary structural performance?
‘ AI-Generated Design Can Be a Hypothesis, Not a Conclusion ‘
Therefore, the value offered by AI is not to eliminate these questions; Expanding the field of solution that we can work on. AI can make the idea visible. Industrial design, on the other hand, questions that idea, develops it and re-establishes it with the requirements of the physical world.
Rather than treating a visual as the representation of a finished product, it can be more useful to consider it a design hypothesis that still needs to be investigated and validated.
At this stage, the visual does not define the final solution. It establishes a design direction—a starting point for form, proportion, surface relationships, details, and product character. Their physical validity is then tested throughout the product development process.
Proportions are evaluated against real components and technical volumes. Surfaces are reconsidered according to materials and manufacturing processes. Part architecture, interfaces, and assembly details are resolved. Ergonomic decisions are validated through use scenarios. Form, function, and manufacturability are tested through prototypes.
In this way, an idea that becomes visible within seconds can evolve through design, engineering, prototyping, and validation into a viable physical product. The final product will not necessarily remain identical to the initial visual. Its form naturally evolves as technical requirements are resolved and design decisions are validated.
The Designer’s Role Is Changing
In the age of AI, it is becoming increasingly difficult to define a designer’s value simply by the ability to create an attractive form or a compelling visual. The time and cost required to generate ideas, variations, and visualizations are falling rapidly.
— However, it is not correct to draw a definite limit on today’s technology.
Given the pace of development, many of the manufacturability challenges we currently consider limitations of AI may soon become parameters that these systems can address directly.
Imagine defining not only the desired form, but also production volume, material, manufacturing process, target cost, components, ergonomic parameters, and performance criteria within a design system. Such a system may eventually generate far more than a concept image. It could produce optimized geometries, part architectures, manufacturing details, and technical outputs based on those inputs. (For a detailed explanation of what can be created with AI in product design, you can check: https://www.ibm.com/think/topics/ai-product-design)
When a surface changes, it could simultaneously evaluate the impact on weight, cost, or structural performance. It could compare different material and manufacturing scenarios and generate technically viable alternatives within seconds. A significant part of the relationships that designers and engineers currently manage across different disciplines and tools may eventually be resolved within the same system.
And when that happens, the designer’s role will inevitably evolve. Because generating thousands of technically viable solutions does not necessarily mean knowing what the right product should be.
Determining which problem is worth solving, understanding what the user truly needs, establishing priorities between conflicting criteria, and recognizing the qualitative difference between two technically valid solutions cannot be explained through optimization alone. Cost can be reduced. Weight can be optimized. Manufacturing can be simplified. Ergonomic targets can be met.
But not every design decision can be reduced to a measurable parameter.
Recognizing a need the user has not yet articulated, understanding the social and physical context in which a product exists, interpreting the character a brand should express, or sometimes choosing to move beyond the direction suggested by the data requires experience, intuition, and design judgment.
Design is not simply about finding the best response to a given set of criteria. It is also about questioning which criteria truly matter—and recognizing what has not yet been defined.
Tools will change. Many technical processes that require specialist expertise today will become increasingly automated. The boundary between what humans and machines can do will continue to shift. And precisely because of this, the human dimension of industrial design may become even more important:
‘ Design is as much about optimizing what can be measured as it is about sensing what cannot yet be measured. Perhaps this is where one of the most critical distinctions between human and machine begins. ‘
- All visuals featured in this article were generated using AI.