Shopify ditches React Native for native apps, crediting AI coding agents

Shopify says it is moving its mobile apps from React Native back to native Swift (iOS) and Kotlin (Android) development, reversing the decision it made in 2020 to adopt React Native in the first place. Back then it chose React Native for three reasons: to stop building the same feature twice, to let developers work across the whole stack even without a mobile background, and to spend less time chasing feature parity between platforms and more time shipping. Shopify says that bet worked: React Native delivered on all three points, and as recently as January 2025 it was still writing that React Native's future looked bright and that it planned to keep investing in the framework. It maintains that React Native remains 'an excellent framework' and that its own React Native apps are fast.
What changed, Shopify says, is coding-agent LLMs. It has used LLMs to build software since 2021, a year before ChatGPT existed by its own reckoning, at first to write code, investigate and fix bugs, and review code. As the models improved, so did the complexity of the work Shopify trusted them with, and by late 2025 agents were no longer just helping engineers write code faster: Shopify says they were capable of making it question whether building a feature twice, once per platform, still meant twice the work. Shopify reevaluated its mobile stack from first principles, prototyped rebuilding core parts of its biggest apps in Swift and Kotlin using LLMs, and says it was surprised by how well that worked. In its account, agents could implement a feature on Android using the existing iOS version as a reference, and vice versa; helped developers ramp up and contribute outside their usual platform; and cut the cost of keeping the two platforms in sync through shared specifications, tests, and review checkpoints.
Shopify is careful to say the underlying cost of native development has not gone away: maintaining two separate codebases, one per platform, is still real work. What changed, in its telling, is that agents can now do enough of the implementation, translation, testing and review that sharing one implementation across platforms is no longer the deciding advantage it was in 2020, while native's own advantages remain: staying closer to each platform's own capabilities and first-party tooling, with fewer framework and dependency layers between the app's code and the platform.
The switch also affects three open-source libraries Shopify built for the React Native ecosystem. React Native Skia keeps Shopify's financial sponsorship through the end of 2026; its creator, William Candillon, plans to fork the project under a new name within the coming months and will keep developing it under that name beyond 2026, after which Shopify's original repository will be archived. FlashList, which Shopify says has become the default way to render high-performance lists in React Native and gets about 2 million downloads a week, will keep receiving fixes for critical compatibility issues from Shopify, which says it is in talks with several companies about taking on its long-term stewardship. Restyle, which Shopify says has a smaller user base than its other libraries, is being archived outright: Shopify will keep it working through the end of 2026 and then stop maintaining it, though it says anyone is welcome to fork it and that it will help with the handover if a team wants to take it on.
Shopify runs several large apps, named Shopify, Shop, Point of Sale and Inbox, that it says millions of merchants and buyers around the world rely on every day. For the migration itself, it chose to rebuild apps from scratch (a 'greenfield' approach) rather than migrate them gradually screen by screen ('brownfield'), which is how it originally adopted React Native. It gives three reasons: LLMs are good at building Swift and Kotlin features using the existing React Native version as a reference; a clean slate lets the team rebuild without the constraints of the old codebase; and its prototypes showed the apps could be rebuilt substantially faster than was possible before coding agents existed. The Shop app was migrated first, going from proof of concept to a fully rebuilt native app published in the app stores in 12 weeks. The Shopify app, the company's biggest with more than 300 screens plus home and lock-screen widgets, an Apple Watch app, complications and Siri Shortcuts, is also being rebuilt natively and is due to ship later this year, with no firmer date given; its remaining apps, including Point of Sale and Inbox, are due to follow only 'soon.'
Shopify says simply pointing an LLM at the React Native codebase and asking it to one-shot the same features natively does not work, even when the model is first asked to gather specs and task files up front: it produces, in its own words, 'a huge amount of unmaintainable code that can't be shipped.' Instead it built an internal system called Helix that works more gradually. A developer points Helix at one screen; Helix reads the React Native version and proposes a sequence of small, ordered checkpoints that can each be reviewed in minutes, then builds checkpoint by checkpoint. Every checkpoint has to prove its behavior with tests, match the running app in a visual review, survive two adversarial code reviewers, and get a human's sign-off before it is committed and the next checkpoint starts; feedback from every review is remembered, so Shopify says the process gets more autonomous as a migration progresses.
Shopify also says agentic control of mobile simulators had been a bottleneck: agents could write code in seconds but needed several minutes to test it, largely because checking an app's state relied on the accessibility tree or on screenshots, and its own tooling for letting agents reproduce and fix bugs autonomously was slow and brittle. Its fix was architectural. Shopify decoupled business logic from the UI so it can run headlessly on a desktop, then exposed it to agents through a command-line interface that lets them inspect app state, navigate between sections and perform actions without touching the UI at all. That cut iteration to milliseconds instead of minutes and, Shopify says, let agents work autonomously for hours at a time. When a simulator is actually needed, the same CLI can drive it remotely by sending commands directly rather than by inspecting the screen layout or accessibility tree, which Shopify says makes end-to-end tests fast as well.
Going forward, Shopify says it will migrate all of its mobile apps to Swift and Kotlin using AI throughout the process, and that every rebuild must match or exceed today's performance, stability, accessibility and product-quality bar rather than simply port the same app into a different language; it plans to measure success through product velocity, app quality, and how much of the work agents can complete on their own. It says it is hiring mobile engineers, infrastructure engineers, and developers working at the intersection of AI and software engineering for the effort, and plans to publish more detail on Helix and what it calls its 'agent-addressable' architecture. The post closes by thanking Meta's React Native team for its stewardship of the framework, thanking William Candillon for creating React Native Skia, thanking Software Mansion for its work on the Reanimated animation library, and crediting the hundreds of Shopify engineers who contributed to adopting React Native, migrating its apps and building shared foundations over the years.
Key facts
- Shopify is reversing its 2020 decision to build mobile apps in React Native, moving them to native Swift (iOS) and Kotlin (Android) because improvements in coding-agent LLMs have cut much of the extra cost that native development used to carry.
- The Shop app already shipped as a fully native app, going from proof of concept to publication in the app stores in 12 weeks; the flagship Shopify app, Shopify's biggest with more than 300 screens plus home and lock-screen widgets, an Apple Watch app, complications and Siri Shortcuts, is underway and due to ship later this year.
- Shopify built an internal system called Helix that migrates one screen at a time through small, reviewable checkpoints: each checkpoint must pass tests, match the running app in a visual review, survive two adversarial code reviewers, and get a human's sign-off before the next one starts.
- Among Shopify's open-source React Native libraries, React Native Skia keeps Shopify's sponsorship through the end of 2026 as creator William Candillon forks it under a new name, FlashList (about 2 million downloads a week) keeps getting critical compatibility fixes while Shopify looks for a new steward, and Restyle is being archived and goes unmaintained at the end of 2026.
- To speed up agent-driven development, Shopify decoupled business logic from the UI so it runs headlessly on a desktop and built a command-line interface that lets agents inspect app state and act without driving a simulator, cutting iteration from minutes to milliseconds and, Shopify says, letting agents work autonomously for hours at a time.
Why it matters
This is a concrete, named example of a company saying that AI coding agents, not new hires or a bigger budget, changed the economics behind a major architecture decision. Cross-platform frameworks such as React Native exist largely because writing and maintaining a separate iOS codebase and a separate Android codebase used to cost more than sharing one codebase across both; Shopify says that cost gap has narrowed enough, now that agents can implement, translate, test and review code across both platforms, that the old calculus no longer holds for its apps and team. Shopify is careful to say this is not a verdict on React Native's quality: it still calls the framework 'excellent' and says its own React Native apps are fast. The story is a real operational data point in the wider question of whether coding agents mainly speed up existing workflows or actually change which workflow makes sense in the first place, coming from a company that says millions of merchants and buyers rely on its apps every day to run their business and buy the products they want.
Who it affects
Most directly, Shopify's own mobile engineers, who move from one shared React Native codebase to platform-specific Swift and Kotlin work guided by Shopify's new Helix migration system. Millions of merchants and buyers who rely daily on Shopify's four largest apps, Shopify, Shop, Point of Sale and Inbox, are the intended beneficiaries once the native rewrites ship. The React Native open-source community is affected too: Shopify is stepping back from three libraries it built and maintained, React Native Skia, FlashList and Restyle, so any team depending on them has to track the transition on its own timeline. More broadly, any engineering organisation currently weighing a cross-platform framework against native development on cost grounds now has a named, public data point to weigh against its own numbers.
How to use it
There is no product or price here; the practical value is a template and a set of dates. Teams considering a similar migration can borrow Shopify's Helix pattern: checkpointed rebuilds gated by automated tests, a visual match against the running app, two adversarial code-review passes and a human sign-off, offered as a way to avoid what Shopify calls 'a huge amount of unmaintainable code' from one-shotting a rewrite with an LLM. Teams that depend on Shopify's React Native libraries have concrete dates to plan around: React Native Skia keeps Shopify's sponsorship through the end of 2026 as creator William Candillon forks it under a new name; FlashList keeps getting critical compatibility fixes while Shopify looks for a new steward; Restyle is archived and goes unmaintained at the end of 2026, open to anyone who wants to fork it. Shopify also says it is hiring mobile engineers, infrastructure engineers, and developers working at the intersection of AI and software engineering for the rewrite.
How solid is it
This is Shopify's own account of its own decision, published in the first person on its engineering blog, not an independent report or an academic study, and no outside party has verified the productivity gains it describes. Some of the underlying facts are concrete and checkable: the Shop app's 12-week run from prototype to a shipped native app, the 300-plus screens in the Shopify app, FlashList's roughly 2 million weekly downloads, and the specific end-of-2026 sunset dates for two libraries. But the central claim, that coding agents specifically closed the cost gap that made React Native worth adopting in 2020, rests on Shopify's own prototyping experience rather than a published benchmark: phrases like 'a fraction of the time' and 'substantially faster' are never attached to a number comparing native and React Native development, so the size of the improvement cannot be checked from the post alone. Shopify does undercut its own case in one place by stating plainly that React Native remains a good framework and that its own React Native apps are fast, which reads as an attempt at a balanced account rather than a one-sided pitch for the reversal.
Risks and caveats
Native development's basic cost, maintaining two separate codebases instead of one, has not disappeared; what changed is how much of that extra work agents now absorb, not the fact that the work exists. Beyond the Shop app, the migration timeline is loose: the flagship Shopify app is due only 'later this year' with no firmer date given, and Point of Sale and Inbox are promised only 'soon.' The switch also carries a cost for the wider React Native ecosystem: two libraries Shopify built and maintained, React Native Skia and Restyle, are being handed off to their creator or wound down entirely as a direct consequence of Shopify's own retreat from the framework, a real loss for other teams that relied on Shopify's stewardship even with forks and handovers on offer. And because this is a self-set, self-reported initiative, the definition of success, product velocity, app quality, and how much agents can do autonomously, is Shopify's own; there is no external benchmark holding the migration to account if timelines slip.
“Native keeps us closer to platform capabilities and first-party tooling, with fewer framework and dependency layers between our code and the platform.”
— Shopify engineering blog