One solo project. Three shipped products. More than 1,500 users.
A developer published the story of building an iPhone app, an Apple Watch companion app, and a marketing landing page entirely with Claudee Code](/tools/claude-code/) - Anthropic's terminal-based AI coding assistant. The apps are live, App Store-approved, and actively used.
The Apple Watch piece is the interesting part. watchOS development is genuinely painful. It involves a separate build target, different UI constraints than iOS, and its own communication layer to sync data with the phone. Most indie developers skip it entirely or treat it as a v2 feature. Getting Claude Code to handle that alongside a full iOS app and a web landing page, within a single project context, suggests the tool is operating well beyond simple code suggestions.
Claude Code is different from in-editor tools like Cursor or Cody. It runs in your terminal and has read and write access to your entire project - it can create files, run commands, and iterate across multiple targets in one session. That whole-project view is what makes multi-platform builds like this feasible for one person without a team.
The 1,500-user milestone is the real signal. Apple's App Store review process rejects apps with crashes, missing entitlements, or broken builds. Shipping something that passes review and attracts real users means the output is production-grade code, not just a throwaway prototype.
For anyone sitting on an app idea, the economics are difficult to argue with. Claude Pro costs $20 per month. Compare that to the $10,000 to $20,000 a typical development agency charges for an iOS app - and that price usually excludes a Watch companion and a marketing site.
The genuine open question is how AI-assisted codebases hold up over time. Debugging a six-month-old project you half-understand is a different challenge than building v1. That test is still ahead for most projects built this way.