Shopify has abandoned its cross-platform mobile development strategy and returned to full native iOS and Android apps, using AI coding agents to handle what would have been a massive duplication of effort. The company announced Thursday that its consumer app, Shop, went from initial concept to a completely native production release in just 12 weeks. The shift reverses a 2020 decision to build mobile features once in React Native rather than writing separate code in Swift and Kotlin, a bet that resonated widely among developers at the time but has now been made obsolete by AI-powered coding tools.
The company is now tackling its far more complex merchant app, which spans over 300 screens and depends heavily on deep integration with iOS platform features. According to the report, coding agents became capable enough by late 2025 that they could examine how a feature functioned on iOS and construct the Android version, or work in the opposite direction. Engineers could also work on platforms outside their expertise because agents handled much of the platform-specific implementation. The productivity shift is dramatic elsewhere in the industry as well: Bun creator Jarred Sumner used 64 parallel Claude instances to port the JavaScript runtime from Zig to Rust, converting roughly a million lines of code in 11 days at an estimated API cost of $165,000, with the existing test suite passing across all six supported platforms before the merge.
"LLMs changed one of the core assumptions behind our 2020 decision," wrote Mustafa Ali, Shopify's director of engineering, in the announcement. The report notes that Shopify doesn't view its React Native bet as a mistake, since the framework delivered exactly what the company required for many years. There was no obvious indication Shopify was preparing to walk away from the approach either; as recently as January 2025, the company was still discussing its future with React Native publicly. The report describes early attempts at simply feeding existing React Native code to an LLM and asking for a native rewrite as producing "slop," with too much code that engineers wouldn't want to maintain even when the team added intermediate steps like writing specs and task files first.
Shopify built an internal system called Helix to manage the agents performing the rewrite, breaking the migration down screen by screen and dividing each one into smaller chunks rather than attempting to recreate everything at once. The system compares results with the existing app as it progresses, while separate agents search for problems in the code before a human approves it, and the system retains that review feedback so issues caught earlier can inform subsequent work. The migration also exposed a testing bottleneck: agents can modify code in seconds, but testing it in a mobile simulator can take twice as long. Shopify worked around this by separating business logic from the UI and letting agents interact with the app through a command-line interface running on a desktop, allowing some checks that once took minutes to complete in milliseconds. The company says it will judge the native apps partly by how much work agents can eventually handle on their own, acknowledging that most codebases weren't built with AI agents in mind but that it's starting to build around them. The economics of major technical decisions are shifting rapidly: developer Simon Willison reported in May meeting an engineer whose company had used coding agents to combine separate iPhone and Android apps into a single React Native app, with the engineer unconcerned about getting locked in since they could always move back to native if needs changed. The same task that would have required a small team multiple quarters a year ago is now an 11-day sprint supervised by one person. For companies willing to build the orchestration layer, the cost of reversing architectural decisions has dropped enough to make platform choices feel less permanent than they once did.