Design strategy
Software Is Getting Cheaper to Copy. Design Is Not.
AI can now attempt to clone major software. Here is what that means for designers, and the questions we should be asking.
In short
- What happened: A developer used AI to build open-source clones of seven Adobe applications. They are early and incomplete, but the attempt itself is the news.
- Why it matters: When anyone can rebuild familiar software, the interface stops being an advantage on its own.
- What it means for designers: Copying screens is getting cheap. Discovering needs, validating workflows, and earning trust are not. That is where design creates value now.
Why this matters for designers
AI-built replicas deliberately reproduce familiar interfaces and workflows. Competitors are targeting not only functionality, but also the familiarity that makes switching easier. [2]
That puts pressure on three things that are often bundled together:
| Advantage | What it is | Can AI copy it? |
|---|---|---|
| Implementation | The code that makes a capability work | Increasingly, yes |
| Product knowledge | Understanding which capabilities and interactions users need | Only what is already visible |
| Operational trust | Confidence that the tool works reliably in consequential workflows | Not by copying |
An AI-built replica can borrow a visible specification from an existing product. That is different from discovering an unmet need, validating a new workflow, or maintaining a dependable product over time. [4, 5]
The strategic question shifts. It becomes less "Can someone copy these screens?" and more "What makes this workflow genuinely dependable and valuable beyond its visible interface?"
When the screens are easy to copy, the value moves to what the screens cannot show.
I have seen this before
I started as a photographer, not a designer.
In the 1980s a high school mentor taught me the darkroom, and that led to a scholarship at the Center for Creative Studies. My craft was chemical. Film, enlargers, trays of developer, and long hours in the dark.
Later I became a biomedical photographer at the University of Maryland Baltimore research hospital. I photographed everything from microscopic specimens to centuries-old mummies. The lab was half darkroom and half digital workstation. I could see which half was growing.
Photoshop was at the center of that shift. I have used it since version 3. The chemical processes I trained for were turning into digital processes.
So I made a choice. I did not try to compete with the software. I went back to school to study how people use it. That led me to Human-Computer Interaction, and to my first wireframe in 1997.
Now the software that changed my first career is being commoditized itself. A single developer with AI can attempt a free version of the suite that once replaced the darkroom.
What I learned the first time still holds. The technology absorbed the process. It did not absorb the judgment. Knowing what makes an image matter, and what a researcher needs to see, stayed with the people. The photographers who did well moved toward that judgment.
That is the move designers need to make now. The questions below are where to start.
Questions designers should be asking
If the screens are the easy part to copy, our work has to go deeper than the screens. These questions help move a conversation from "what should this look like" to "what makes this worth building."
About product knowledge
Do we understand the need, or are we copying a solution?
- What did we learn about this workflow that is not visible in any competitor's interface?
- Which assumptions have we validated with real users, and which did we inherit?
- If someone cloned our product tomorrow, what would they still not know?
About operational trust
Does this work when it matters, not just in a demo?
- Where do mistakes in this workflow carry real consequences: money, legal standing, safety, reputation?
- How do we know it works reliably over time, not just on the first try?
- When it fails, does the person using it know, and can they recover?
About implementation
Are we spending our time where copying is cheap?
- Is this request asking for screens, or for a solved problem?
- If a team could build only the part they need, what would they still need design for?
- Which parts of our process could AI do, and what does that free us to do instead?
To bring to product partners
Is design invited in before the answer is decided?
- What problem are we solving, and how will we know it is solved?
- Is discovery funded, or only delivery?
- Who owns understanding the user over time, after launch?
What happened
The project is ArtCraft. Its developer, Brandon Thomas, says he used AI to create open-source replacements for seven Adobe applications. Reporting describes them as "clean-room" reimplementations: attempts to reproduce functionality without copying Adobe's proprietary code or assets. [2, 3]
A qualification. "Someone replicated the entire Creative Suite" is stronger than the evidence supports.
- The applications are early alpha builds.
- Testers describe them as familiar and functional in places, but buggy and incomplete.
- Full feature parity is the developer's goal, not a demonstrated result. [3, 4, 5]
The real development is that a small project can now attempt a much broader competing suite than previously seemed practical.
Why copying is getting cheaper
The underlying concept is commoditization: capabilities that once required scarce expertise and investment become widely available and harder to charge a premium for. [1]
AI-assisted building creates three pressures at once:
| Pressure | What it means |
|---|---|
| 1. Smaller teams can build more | The same product takes fewer hours, or the same team ships more. [1] |
| 2. More competitors can enter | Credible alternatives come from people who previously lacked the resources to try. ArtCraft is an example. [1] |
| 3. Customers build only what they need | A team generates the reporting, approval, or data workflow it needs instead of buying a large application. This one is still a possibility, not yet a demonstrated trend. |
The third pressure is the one to watch in enterprise settings. The alternative to a paid product may be "we made only the part we need."
The part that cannot be copied
The ArtCraft clones are rough today. They will get better. The screens will keep getting easier to reproduce.
So the question for designers is not whether our interfaces can be copied. They can. The question is whether we are doing the part of the work that cannot be.