Early on, I learned something that took me longer to accept than it should have: working code is only half the job. If the software looks like a developer designed it — and by that I mean a wall of buttons, three nested dropdowns, and a form that requires you to read the manual — clients will not trust it. They will not use it comfortably. And sometimes they will blame it for being broken when it isn't.
Developers, myself included, tend to be practical people. We think in terms of function first. The result, when we design for ourselves, is software that a developer finds perfectly obvious and a normal human finds disorienting. I learned that lesson once. I did not need to learn it twice.
Before the talent existed
The problem was finding someone who could fix it. This was well before freelancing platforms became the default way to hire talent. Before Upwork, before Behance, before design became something you could search for by portfolio and hourly rate. Finding a decent UI/UX designer in the region meant knowing the right people — or knowing people who knew the right people.
I contracted interior designers more than once. Not as a joke — as a genuine workaround. They understood color relationships, spatial hierarchy, what makes a layout feel balanced versus cluttered. The translation from physical to digital was imperfect, but it was better than leaving the visual layer to me. Some of it actually worked.
Eventually, I found good UI/UX designers — people who understood both the aesthetic and the functional side of software. I still work with some of them today. That part of the story has a reasonable ending.
The market did something weird
Here is where it gets strange. Over the past decade, the number of people calling themselves UI/UX designers grew substantially. Portfolio platforms, design bootcamps, YouTube courses — the supply expanded faster than most technical disciplines. Normally, when supply outpaces demand, prices fall.
That did not happen.
The good designers got more expensive. Significantly more expensive. And they became more selective — declining projects that didn't fit their portfolio direction, or that they felt they couldn't deliver on, or simply that didn't interest them. A senior designer turning down work is a reasonable thing to do. When it becomes the norm, and the rates for the ones who do say yes keep climbing, the economics start to feel disconnected from the underlying math.
The irony is that most software design is not experimental art. People prefer patterns they already recognize. Put the navigation where they expect it. Make the buttons look like buttons. Let the workflow follow the way they already think. There is genuine skill in executing that well — but it is not the kind of skill that should require months of budget to access.
The best designs don't make you notice the design. They make you notice that the software works.
What AI actually delivers
I started testing AI on design tasks with the same skepticism I bring to anything that arrives with this much noise around it. What I found surprised me — not because it was perfect, but because it was genuinely usable.
AI can produce interfaces that follow UI/UX guidelines. It understands visual hierarchy, spacing systems, color contrast, component consistency. Given the right direction, it generates something that a real designer would recognize as competent work. Not always on the first attempt — it rarely comes straight out of the box. But with a few rounds of back-and-forth, with clear feedback and specific corrections, you get somewhere close to the target.
I talked about this with several developers I work with. The consensus was not that AI design was revolutionary — it was that the bar for "good enough" had shifted. For a project where the design needs to be clean, functional, and consistent with established patterns, AI gets you there. It does not replace the designer solving a genuinely novel design problem. But those projects were never the majority.
Where developers are right now
Most of the developers I spoke to said the same thing: designers had become overpriced for what was being asked. Not for truly complex, original design work — for standard software interfaces that needed to look professional and function correctly. For that category of work, the price-to-output ratio had drifted into territory that was hard to justify.
So they are experimenting. Not abandoning designers entirely, but testing AI on new projects — particularly smaller ones, internal tools, MVPs — and finding the results acceptable. Some are finding them better than acceptable.
The designers who will continue to command high rates are the ones who deliver things AI cannot yet reliably produce: the considered design decision, the deliberate departure from convention, the interface built for a genuinely unusual workflow. That work exists and will continue to exist.
The rest of it — the clean, professional, familiar software interface — has a new competitor. And the competitor doesn't negotiate its rates.
