$3.7K in 28 Days From a Solo App Built With Claudee Code](/tools/claude-code/)
$3,700 in 28 days. That's what one developer earned from an iPhone and Apple Watch app they built using Claude Code - and it's the kind of number that makes other developers stop scrolling.
The app crossed 3,300 users, generating real revenue from what appears to be a solo build. Claude Code, for those who haven't used it, is Anthropic's command-line coding tool that can read your codebase, write code, run tests, and fix bugs with minimal supervision. It's aimed at developers, but the barrier to shipping something has dropped considerably for anyone willing to learn the basics.
Why Apple Platforms Used to Be a Hard Stop
Building for iOS and watchOS has historically demanded a steep on-ramp: Swift as a language, Xcode as the development environment, Apple's interface guidelines, App Store review rules, and the specifics of WatchConnectivity for syncing data between iPhone and Apple Watch. That's a significant amount of platform-specific knowledge to absorb before you can even test whether your app idea has legs.
Claude Code compresses that curve. Not because it outperforms a senior Swift engineer on complex problems - it doesn't - but because it handles boilerplate, looks up obscure APIs, and debugs compilation errors without requiring the developer to understand every detail. Someone with general programming experience can now reach a shippable iOS build faster than at any point before.
Revenue Is a Cleaner Signal Than Benchmarks
Eval scores on coding leaderboards are one thing. An app making money in the App Store is another. Passing Apple's review process (notoriously picky about crashes, privacy labels, and UI consistency), acquiring thousands of real users, and converting enough of them to paid customers - that's a sequence Claude Code visibly supported.
This doesn't mean the tool replaces a dedicated mobile developer for anything complex. Long-term maintenance, performance tuning, and handling edge cases in production are still genuinely hard problems. But for validating an idea quickly and getting to a v1? The evidence is there.
The real constraint hasn't shifted: Claude Code can build the thing, but it can't decide what to build. That judgment call - figuring out what people will actually pay for - still belongs entirely to the developer.