Why We Went Native

Mint started as an Electron app. We picked Electron for practical reasons: it runs cross-platform, we already knew the web stack, and we could prototype fast. The prototype worked well, and we still ended up deleting it. ### What Was Wrong With the Prototype The Electron version had a smooth mind-map canvas, drag-and-drop that felt natural, and clean code. It made a good demo. It didn't hold up as something you'd use every day. Three things made the decision for us: 1. **280 MB on disk.** That's a lot for a file organizer. We couldn't get past a cleanup tool being bigger than most of the files it manages. 2. **4-second launch time.** A utility should open right away. Four seconds is long enough to forget why you opened it and close the window. 3. **Sandboxed file system access.** Every operation needed a workaround. Moving files across volumes needed temporary permissions, and watching directories for changes was unreliable. We spent our time fighting the OS. ### The Question That Decided It We asked ourselves whether we'd keep a 280 MB file organizer running in the background. We wouldn't, and we didn't want to ship something we wouldn't use ourselves. ### What We Got From Going Native Here's what the SwiftUI rewrite gave us (the size is measured on Mint 1.0.33, the version you can download now): - **A 50 MB download.** That's version 1.0.33. Mint does a lot more than organize files now, with Disk, Memory and Uninstall in the same app, and the download is still a fraction of the 280 MB Electron prototype. - **Direct file system access.** The version you download from our site isn't sandboxed, so it works with your files directly. The Mac App Store version is sandboxed and works in the folders you give it. - **Folder-scoped automatic organization.** Automation only runs in folders you've added, and you set each one to Manually, Daily at a set time, or As files arrive. - **Apple Vision and native language analysis.** OCR and on-device content analysis, with no pile of native modules or fragile bridges to maintain. We care about this because you notice it every day. A utility that's slow to open and heavy on memory feels bolted on to your Mac. ### What Transferred We didn't lose everything. The mind-map layout algorithm (pure math: force simulation and collision detection) translated directly from TypeScript to Swift. So did the data model and the design decisions. We'd already worked out the visual hierarchy, the interaction patterns and the information architecture. The rendering layer and state management didn't carry over, so we rewrote those from scratch. We later cut the mind-map canvas itself, so you won't find it in Mint today. ### Looking Back We don't think the Electron prototype was wasted time. It was a proof of concept, and it taught us what the product needed to be. The native app came out better because we'd already made our design mistakes in the prototype. The code we threw away worked. It just wasn't good enough for an app people would open every day.