A Mac App Took 11 Days and Zero Handwritten Code With Claude — Then Apple Approved It

A Mac app went from first prompt to a paid product on Apple's App Store in 11 days, with no code written by hand. The developer behind it used Anthropic's Claude to generate the app and shepherd it through Apple's review process, which is often the slowest and most unpredictable part of shipping software for the Mac.

The account, published by ZDNet, describes an end-to-end run: idea, implementation, and App Store approval, all inside a window that would normally cover little more than initial setup for a conventional project. That timeline is the real headline here. Apple's Mac App Store review has a reputation for rejections over metadata, sandboxing rules, and entitlement mismatches, so clearing it on a first pass after an 11-day build is the part that will interest anyone who has shipped there.

This fits a pattern that has been building through 2025: large language models are absorbing the mechanical parts of app development — boilerplate, UI scaffolding, API glue, error handling — leaving the human to specify behavior and judge output. Claude has become a common choice for that work because of its strength at long-context code reasoning and its ability to hold a whole small codebase in view. What is different in this case is not the coding assistance itself but the scope: the claim runs all the way to a purchasable product, not a prototype or a demo.

The economics are the disruptive part. An indie developer who once needed weeks of evenings and a tolerance for Apple's documentation can now test a product idea in days, at the cost of a subscription. That lowers the bar for the long tail of small, single-purpose Mac utilities that large studios ignore because the revenue never justifies the effort. It also means the App Store's supply of low-complexity apps is likely to rise, which cuts both ways for anyone trying to stand out in a crowded category.

It is worth being precise about what the 11-day figure does and does not prove. The developer still made the product decisions, defined the scope, and handled submission, and one successful build says nothing about how these projects behave at scale or under maintenance. AI-generated code that passes review can still carry subtle bugs, security gaps, or licensing questions that surface months later. Apple's guidelines also do not exempt machine-written apps from any of its rules, so the review bar is unchanged — only the speed of reaching it has.

What to watch next is whether this becomes routine or stays an anecdote. The signals will show up in App Store submission volumes, in whether Apple adjusts review or disclosure requirements for AI-generated software, and in how many of these fast-built apps survive their first round of user bug reports. The interesting question is not whether a model can write a Mac app in 11 days. It is whether the resulting product is worth paying for a year later.