)
)
TL;DR
AI made building apps a doddle. It did absolutely nothing for the humans on the other end.
More releases, same attention spans, same instinct to bail the second something feels clunky. Here's what that means in practice:
AI defaults to the most common design patterns — which means your app risks looking exactly like everyone else's
The first 90 seconds of onboarding still make or break whether someone stays
Error states, trust signals, and iteration flows need a proper human brain behind them — not a default component library
The friction points are where the real design work lives, not the easy screens
UX is now what decides whether your shiny new build gets opened twice or never again. Design accordingly!
UX in the Age of AI
Let's get the boring bit out of the way first. Artificial Intelligence has absolutely cracked open how apps get built. We're not here to clutch our pearls about it. We use AI tools at Cube every single day, and frankly, they're brilliant. The barrier to shipping software has never been lower, which means more sharp ideas from more sharp people are finally getting their shot. Mission-led orgs that used to wait years for a slot now have one live by the weekend.
We love to see this!
But let’s just address the elephant in the algorithm: your users didn't get the memo. They're still scrolling with the same thumbs, the same attention spans, and the same instinct to bounce the second your app makes them feel a bit thick. The way apps get built has changed. The humans on the other end have not. Not even slightly.
Which is exactly why User Experience (UX) is having a moment. Or, more accurately, why it's what quietly (but essentially) decides which of all these new apps actually survive.
The Apps Are Multiplying. The Users Aren't.
The receipts back this up. The National Bureau of Economic Research (NBER) tracked more than 100,000 developers using AI tools in 2026, and AI did exactly what it says on the tin: code output up as much as 180%. But that's where the good news peaks. Follow the code towards a living, breathing user and it bleeds out at every step. 180% more code. 50% more finished projects. 30% more things that actually ship. Turns out the slow part was never the typing.
The app stores have clocked it too. Google Play has quietly cut its catalogue almost in half, down from around 3.4 million apps to roughly 1.8 million, by chucking the junk and tightening what's allowed through the door. At the same time, the flood keeps coming: new app releases were up 60% on last year in the first three months of 2026, and more than doubled by April.
More apps arriving, fewer making the cut. Which leaves you with two bars heading in opposite directions. Just getting something built and live? That bar's on the floor. Being good enough to survive the cull, get found, and actually keep people? That one climbs higher by the day.
So the floor is full. What wins is everything that happens above it. And the most essential part of that, the thing users feel the second they tap the icon, is the user experience.
(We've done a deep dive on the "anyone can make an app, but not everyone can make a great one" thread in our wider blog on AI and app quality Check’r out!
)
Why AI Builds Often Look Fine and Feel Forgettable
Something that's been rattling around the studio, courtesy of our Head of Product Design, James Marriott. When you ask an AI tool to design or build a screen, it tends to reach for the most well-worn component libraries it already knows. Default cards. Default modals. Default everything, basically the status quo of what's already out there. It's quick, it's confident, and it's about as memorable as a magnolia hallway.
"When the prompt is top-level, that's when all of the platforms tend to use very default patterns. You have to selectively dedicate time to come up with more bespoke interactions, patterns, and layouts." James Marriott, Head of Product Design, 3 Sided Cube
Translation: if you let AI freestyle the UX, you'll end up with something that looks like everyone else's app. Which, by definition, is not the app anyone's making space for on their home screen.
This isn't a problem if you're building a bog-standard internal admin panel nobody's meant to feel anything about. It is very much a problem if the whole point is to get someone to use, trust, return to, or pay for what you've made.
The Bits That Still Need a Proper Brain Behind Them
You don't need a bespoke design magic wand on every screen. You need it on the parts that actually decide whether someone stays or goes. The friction points. The conversion moments. The "this is where you lose them" moments.
A few examples of where AI's defaults consistently fall short without serious human direction:
Onboarding that doesn't feel like a hostage situation. The first 90 seconds are make-or-break, and "default empty state with a wall of placeholder text" is not a strategy, it's a goodbye.
Trust signals when AI is doing the thinking. Probabilistic outputs need transparent design. Where did this answer come from? How confident is it? Can I push back? Default chat patterns don't even try to answer any of this, and your users will notice.
Iteration that doesn't punish curiosity. Users will refine, retry, reshape, and change their mind three times. If your interface treats every retry as a full reset, you've lost them by attempt two.
Error states with a human in them. The AI will get it wrong. Sometimes hilariously. Good UX makes recovery feel like part of the journey, not like a brick wall with no door.
This is the stuff that quietly decides whether your app gets the "used it once" review or the "I open this every day" one.
How We Think About It at Cube
We've been doing this since 2009 and we have absolutely tried and tested feelings about it. The principles haven't shifted, even with AI in the mix:
Users first, always. Tech serves the human. Not the other way round, ever.
The friction points get the love. If a screen is doing something users will hesitate over, that's the screen that earns proper attention.
Test it on actual people. Not your dev team. Not the AI. People who've never seen the thing and won't be nice about it.
Defaults are a starting point, not the finish line. If your product looks like every other product, you've stopped designing and started photocopying.
)
Our Product Design team is now leaning into AI to take care of the predictable, repetitive bits, which frees them up to invest properly in the interactions that actually move the needle. Less time wrestling with default patterns. More time on the moments that make your app feel like something. Same discipline, sharper tools.
So What Now?
Build with AI. Ship faster. Lean into the tools, we do, every single day. But pretty please don't confuse "I shipped an app" with "I made something people want to use." Those two sentences are doing very different jobs, and only one of them grows a user base.
If you've already prototyped something and you're wondering whether it's ready for real humans, we've written about that exact moment in You Vibe-Coded Something, Now What?
And if you're at the front end of all this and wondering how to scope the thing properly in the first place, our guide to scoping without building on quicksand has you covered.
The agencies, founders, and teams who'll win this era aren't the ones with the slickest AI stack. They're the ones who remember that on the other side of every interface is a human who'd really rather not have to think about it.
So build the thing. Just remember who it's for.
Published on 17 June 2026, last updated on 17 June 2026