iOS for Web: Run and Edit Native UIKit Apps in the Browser, Simulator Optional
Running native iOS UIs in the browser without a device cuts the UI iteration feedback loop from minutes to under a second, and lets AI agents modify real UIKit code directly. It also signals that the simulator may not remain the only preview surface for mobile development.
Dessn's iOS for Web claims to turn native Swift and Objective-C projects into browser-rendered, interactive UIs. It operates as a code-based Figma: load a real Xcode project, and the UI appears in the browser with working navigation, gestures, and constraint layout. Developers can duplicate the same project into multiple independent design versions, tweak each one, and see results in under a second, without going through the full IPA build, link, and install cycle.
The implementation is closed-source, but Dessn says it relies on a Rust compiler and renderer. The author speculates the actual pipeline uses WebAssembly plus an intermediate representation, possibly with a custom canvas backed by WebGPU. The core problem is mapping UIKit semantics to the browser: event delivery, state storage, and routing have to be reimplemented for every UIViewController interaction.
Because the tool syncs UI changes back to source code, a coding agent can edit the real project in the browser and have those edits land directly in the codebase. This inverts the usual Figma-first flow, where design is translated into code later.
Treating source code as the single source of truth in a design canvas inverts the Figma-to-code workflow, which could shift how AI coding agents prototype user interfaces.
Closed source leaves the runtime mapping unverified, so the demo's breadth is hard to confirm beyond the shown interactions.
Rewriting UIKit semantics for the browser is the hard part; simple buttons are easy, but real apps rely on gesture recognizers, constraint layouts, and third-party components that demand a full runtime.
Awe at AI's capability collides with skepticism about the closed-source approach. AI and WASM are credited as the enablers, yet the lack of open source is seen as a limit on reach. A question about Swift cross-platform acceleration also surfaces.
AI is now so incredibly powerful that the barrier to programming languages has almost disappeared.
Without open source, it's destined not to go far.
So will Swift cross-platform development eventually become faster?