Clients now arrive with a working Lovable prototype before the first design conversation happens, and that's quietly breaking the brief, and I often hate it.
Muhammad Abdullah Ali
With an abundance of AI tools like Claude Design, Loveable, Aura etc, clients have now started to show up with a working AI prototype of a website or product. But why I have to hate this? Isn’t a good thing when there is more clarity on design from client’s perspective and less time spent on design directions? That’s what I am gonna talk about.
Where It Quietly Goes Wrong
Here's the problem nobody's saying out loud. Most clients don't know how to prompt these tools for anything specific to their brand, so what comes out looks like ten thousand other AI-generated screens. Same section order, same layout logic, same rounded card grid. The tool never knew their actual users. It never knew what made their brand different from the competitor down the street. It just produced something that looked finished, which is a completely different thing from being right.
Now the problem is that AI-generated design becomes a standardized ceiling in the client’s mind. He just needs something like that but a little polished, according to the brand guidelines, and yeah, it pretty much works out for their aesthetic sense.
Now what this thing leads to?
- The founder arrives with a reference instead of a problem.
- The designer ends up working inside someone else's assumptions and pre-decided layout layout.
- The brand identity conversation never happens, because there's already a template sitting there.
- Creative direction gets decided before anyone's actually understood the problem.
- There's a quiet psychological anchor now, and it's hard to talk a client out of something they already built with their own hands.
This hits hardest on fast timelines. When there's no room to slow down, the AI reference becomes the path of least resistance for everyone in the room. The back and forth that should happen gets skipped, not because nobody wanted it, but because nobody had the bandwidth to fight for it under a deadline. And the person who actually loses in that trade isn't the designer. It's the business owner who now has a brand that looks like everyone else's AI output.
Suggestions & Key Takeaway
I won't take a client's AI prototype and just execute it without having the harder conversation first. Not because I enjoy being difficult. Because that reference might be the exact thing standing between their brand and the outcome they actually hired me for. Now I don’t say that I’m prone to this kind of work, but I always try to push back, whenever such a brief lands, not ignoring the constraints and timeline.
The approach clients should be taking or you as a designer while pushing back is to understand what the problem you are trying to solve. Understand where you both are stuck at (maybe a style or flow) and what can genuinely benefit the business. Now push back on it and give valuable suggestions if you are a designer or keep room for new ideas if you are the client. The tools aren't the issue. Lovable, v0, Stitch, Claude Design all of it, genuinely useful for feeling out an idea fast, and I use versions of this stack myself for prototyping. The issue is mistaking a fast first draft for a finished creative decision.
What works is you design one concept as per the direction, and put in some extra effort and deliver another concept which includes your thinking and suggestions. Its a win-win for both but this mostly depends on your position, timeline, and budget.
