How to Build a Mobile App With Claude Code (No App Store Team Required)

The short answer

Yes, you can build a working mobile app with Claude Code without hiring a mobile developer. The practical path is React Native with Expo: one codebase that runs on both iPhone and Android, which Claude Code can write and which you can preview on your own phone in minutes using Expo's free preview app, no App Store submission required to test it. Building the app and publishing it are two different jobs - this post covers both, including what App Store and Google Play actually charge before you go live.

Why I built one instead of hiring a mobile developer

I had a simple idea: a members-only tool that needed to live on a phone, not a browser tab, because the people using it would be checking it standing up between tasks, not sitting at a desk. Hiring a mobile developer for a first version of an idea I had not validated yet felt backwards - I wanted to know the idea worked before paying someone to build it properly.

So I asked Claude Code to build the smallest working version instead: three screens, one core action, running on my own phone. Not a polished release, a real test of whether the idea held up once it was in my hand instead of in my head. [How to build an MVP in a weekend](/blog/how-to-build-an-mvp-in-a-weekend) covers the same instinct applied to a web product - the mobile version follows the same logic, just with a phone instead of a browser as the proof.

Why React Native with Expo, specifically

A native iPhone app and a native Android app are, technically, two separate codebases written in two different languages. React Native collapses that into one codebase that runs on both, which matters enormously when you are not a mobile developer and do not want to maintain two versions of the same idea. Expo sits on top of React Native and removes the part that used to require Xcode or Android Studio installed and configured correctly - you can preview a build on your own phone through Expo's free app almost immediately after Claude Code writes the first version.

How I built it

1. Wrote down the one thing the app had to do, in one sentence, before opening Claude Code at all. Everything that did not serve that sentence got cut from the first version.
2. Described the app to Claude Code in plain language and asked for a React Native project built with Expo, with the three screens the core action actually needed.
3. Installed the Expo Go app on my own phone and scanned the QR code Claude Code's build produced, which put a working preview on my phone in minutes, not after a submission or a review.
4. Used the app for real, on my actual phone, for a few days before touching another feature - screen-sized taps and one-handed use surface problems a desktop preview never will.
5. Asked Claude Code to fix the specific friction I hit using it for real, one fix at a time, instead of listing every possible improvement up front.

What broke, and what App Store submission actually requires

My first version worked perfectly in the Expo preview and then behaved differently the first time I tried a real build for a physical device outside of preview mode - a permissions prompt I had not accounted for stopped the flow cold. That gap between preview and a real build is common enough that I now budget time for it on every mobile project rather than treating it as a surprise.

Publishing publicly is a separate step with real, fixed costs. An individual Apple Developer Program membership is $99 a year and is required before you can submit anything to the App Store; Apple then reviews every submission, which can take anywhere from under a day to several, and can reject an app that does not meet its guidelines. Google Play is a one-time $25 listing fee per developer account, with a lighter review process than Apple's. Neither of these is optional if the goal is a public listing - both are separate from anything Claude Code itself charges you to build the app.

[How to measure the ROI of AI in your business](/blog/how-to-measure-the-roi-of-ai-in-your-business) is worth running before you pay either fee - confirm the app is solving something real for a real user first, since the $99 and the $25 are sunk costs the moment you submit.

The One-Flow Rule, and when a real mobile team wins

Judge a first mobile build against the One-Flow Rule: it is working the moment one complete task goes from start to finish on a real phone in your hand, not when every screen you originally imagined exists. A three-screen app that reliably does the one thing it promises beats a twelve-screen app where half the flows are still broken - the second one is not closer to done, it is just wider.

This is also where the honest limit sits. If the app needs to work fully offline with data that syncs later, use hardware most consumer phones don't expose cleanly, or needs the kind of performance a game or a camera-heavy app demands, a mobile developer who has already solved those specific problems is worth paying for. The Claude Code path here is for a first working version of a straightforward idea, not a replacement for a real mobile team once the app has to do something genuinely hard.

How I coach members through this build

When a member tells me their idea needs to be a phone app, the first question is always whether it actually does, or whether a website that works well on mobile would do the same job without App Store fees and review delays at all. Plenty of ideas survive as a mobile-friendly website. The ones that genuinely need to be an app - because they use the camera, need to work with no signal, or need to send a push notification - are the ones worth this build.

[How to build an app with AI, no code](/blog/how-to-build-an-app-with-ai-no-code) covers the wider decision between a no-code builder and a Claude Code build - the mobile version of that same choice comes down to whether you want to own the code afterward or just want the app live fastest.