Why AI could make native app development great again

追加: 2026-05-14

[00:00:00]I think AR will have a big impact on how we develop apps. Anyone who's just a little bit technical can in the end tell pretty fast whether a mobile app is a native one or written in a cross-platform framework like Flutter or React. And if we ignore the entire development part of having to write code twice, then native apps are just better.

[00:00:16]That's what nobody really denies. So, they are more fluent, they are more performant, they just feel much closer to the platform, it's easier to talk to system APIs, and so on. But, unfortunately, we can't ignore the entire development part. And it's quite rare that you need an app only for one platform. But, no, typically, if you have an app you're serious about, you want to release it for both iOS and Android. And that is where native development can sometimes be the right choice, but where you will also very likely start thinking about building it in a cross-platform framework. Since even as a native Android developer myself, it's very hard to ignore the advantage that with a cross-platform app, you can simply cut your team and development costs by up to 50%. It's just too big of an advantage in many cases. Especially when the app you're building is mainly data-driven and doesn't really need a deep system integration. And that advantage is so big because for a somewhat serious app, this saves you tens of thousands of lines of code that you would otherwise have to write in addition because you need the same app twice. But, with all those AI tools and agents writing the code for existing specs isn't the bottleneck anymore. To be fair, writing code has never been the main bottleneck of development in general, but still in native mobile development, massive time goes into writing the same code twice in a slightly different way. Different SDKs, maybe different language, but the overall functionings and the big bottleneck of modern software development is still what it has always been, coming up with a good detailed specification. Knowing what makes an app successful and useful, knowing your users, knowing your stakeholders, knowing what good UX is, defining edge cases, knowing where performance matters and where it doesn't so much. Because of this foundation of defining what means success for a given app is brittle, then iterating on that with AI will just give you an app that isn't useful for solving the original problem. And having the solid foundation of a good detailed specification is completely independent of whether you build two native apps, a cross-platform app, or two native apps and a website. So, cross-platform frameworks don't change that. But, what AI changes is that writing large amounts of code is not taking that long anymore.

[00:02:17]And what that means is if you have that solid specification, then you can also get to two native apps reasonably fast, or at least by far not taking twice as long anymore. Sure, there will be a little overhead compared to building an app with the help of AI and a cross-platform framework, but that overhead is much, much smaller than it was before. And it's so small that I would personally seriously consider building an app as two native apps rather than using a cross-platform framework if I was serious about building that app, because the advantages of fluent native apps suddenly feel prominent again as the scale of that right code twice disadvantage shrinks. And there is actually also a middle ground, which most of you already know, and that is Kotlin Multiplatform. And this is really where I think that Kotlin Multiplatform is absolutely superior when compared with any other multiplatform or cross-platform framework, because it lets us choose to write the Android UI in Compose, to write the iOS UI in SwiftUI, and share the rest. That will give us a true native experience while sharing at least 50% of the app for API calls, the DB schema, and so on. So, using Kotlin Multiplatform cuts this right code twice disadvantage even more since the technology already works in a way to give us two native apps. If you use Compose Multiplatform and share the UI in a Kotlin Multiplatform project, you don't really have this native feel, because then the UI is not native anymore. But, having the option to write UI boilerplate so much faster with AI for existing specification, for existing mockups, writing separate UI code becomes much more attractive, and you get the full native UI performance. So, if AI assisted development is really something you want to get started with, then the best thing to do is to watch my Cloud Code Crash Course here and download my cloud code skills for Android architecture. So, really skills that make your agent actually stick to my preferred coding standards. It's a completely free download. You can find it in this video's description. But, I'm also really curious to hear your views about this. As most of my followers here are probably native Android developers, does AI excite you? Does it make you worry more? Let's have a little chat in the comments. Thanks for watching. I will see you back in the next video.

[00:04:13]Have an amazing rest of your week.

[00:04:14]Bye-bye.

[00:04:16]>> [music] [music]