AI stopped being the selling point and became the plumbing
I remember when slapping "AI-powered" on your app description was enough to get people excited. Those days are gone.
Now, if your app doesn't do something smart — predict what a user needs next, process a photo on-device, understand natural language input — it just feels behind. Not broken. Just... behind. Like an app without dark mode in 2021.
The shift that genuinely surprised me was how much of this now runs locally. I spent years assuming on-device ML was a niche thing for big teams with hardware engineers on staff. Now I'm shipping it in side projects. Apple's Core ML and Google's ML Kit have gotten good enough that I barely think about it. No server round-trips, no latency spikes, no explaining to users why their data left their phone.
Prompt engineering has also become a real skill on our team. We have a backend developer who basically lives in system prompts now. Six months ago that would've sounded strange. Today it's just Tuesday.
The cross-platform debate got boring (in a good way)
Look, I've had the native-vs-Flutter argument more times than I can count. At this point I'm tired of it, and I think most experienced developers are too.
Here's where I've landed: Flutter is my default. React Native is my default if the team already knows JavaScript. Native Swift or Kotlin is for when I hit something I genuinely can't do any other way — which, in 2024 and 2025, happened less than I expected.
The performance gap that people used to cite as a dealbreaker? For most apps, it's just not there anymore. I shipped a pretty animation-heavy Flutter app last year and nobody on the team or among users brought up performance once. That wouldn't have been true three years ago.
Where I still go native: anything deep in the camera pipeline, ARKit experiences, widgets with real-time sensor data. But that's maybe 20% of what I build.
Foldables are real now, and they changed how I think about layouts
I resisted taking foldables seriously for longer than I should admit. They felt like a gimmick — a product looking for a use case.
Then I started using one as my daily driver for a few months and my thinking changed pretty fast. The users who have them are power users. They open apps expecting to do more. They multitask. They use split-screen constantly.
Designing for adaptive layouts used to be something I'd do if time allowed. Now it's in my checklist from day one. Not because every user has a foldable, but because thinking about adaptive layouts forces you to make better decisions about information hierarchy across the board. It's made my regular phone layouts better too, which I didn't expect.
Voice just kind of snuck up on me
I didn't consciously decide to add voice support to more of my apps. It sort of happened because users started expecting it.
I noticed it first in user feedback — comments like "wish I could just tell it what I want" showing up more often. Then I started watching people use their phones on the train and realized how often they're talking to their apps, not tapping.
Voice isn't replacing touch. But it's becoming a real second input method, especially for anything that involves searching, filtering, or filling in fields. Apps that pretend it doesn't exist are starting to feel like apps that pretended touch screens didn't exist in 2009. Not wrong exactly, just not reading the room.
Privacy pressure is real and honestly I think it's good
The App Store review process has gotten more intense. Privacy nutrition labels actually matter now because users look at them. I've had friends — non-technical people — tell me they deleted apps because of what the privacy label said.
This used to feel like a compliance headache. I've come around on it. Processing things on-device, asking for only the permissions you actually need, being upfront about what you collect — all of that builds trust in a way that shows up in retention metrics eventually. Users notice when an app doesn't feel exploitative, even if they can't articulate why.
The stuff nobody tells you
All the technical shifts are real. But the thing I'd actually tell a developer just getting into mobile apps in 2026 is this: the craft of building something people want to use hasn't changed at all.
Users still bail in the first 30 seconds if onboarding is confusing. A slow startup time still feels like disrespect. An app that solves a fake problem still fails no matter how well it's built.
I've shipped things with cutting-edge ML that flopped, and I've shipped simple utilities with no AI whatsoever that people have used every day for years. The technology is more interesting than ever. But it's still just the means to an end.
Build something worth someone's attention. The rest is details.
A lot of modern mobile experiences now depend heavily on AI-powered features , whether that's recommendation systems, voice interaction, or intelligent automation running directly on-device.
Maintaining consistency across mobile platforms becomes significantly easier with a strong scalable design system that teams can reuse across products and releases.
As apps process more sensitive user data, mobile teams are also adopting ideas from zero trust security models to improve authentication, access control, and device-level protection.