{"id":47176,"date":"2025-03-24T16:38:51","date_gmt":"2025-03-24T20:38:51","guid":{"rendered":"https:\/\/mjtsai.com\/blog\/?p=47176"},"modified":"2025-07-14T11:11:55","modified_gmt":"2025-07-14T15:11:55","slug":"tim-dont-kill-my-vibe","status":"publish","type":"post","link":"https:\/\/mjtsai.com\/blog\/2025\/03\/24\/tim-dont-kill-my-vibe\/","title":{"rendered":"Tim, Don’t Kill My Vibe"},"content":{"rendered":"

Bryan Irace<\/a>:<\/p>\n

It’s one thing for Apple’s AI product offerings<\/em> to be non-competitive. Perhaps even worse is that as Apple stands still, software development is moving forward faster than ever before. Like it or not, LLMs—both through general chat interfaces and purpose-built developer tools—have meaningfully increased the rate at which new software can be produced. And they’ve done so both by making skilled developers more productive while also lowering the bar for less-experienced participants.<\/p>

Barring a sharp correction, Apple looks increasingly likely to miss out on a generation of developers conditioned to first reach for tools like Cursor<\/a>, Replit<\/a>, or v0<\/a>—especially as Apple’s own AI tooling remains notably absent<\/a>. This goes well beyond enabling new entrants to “vibe code<\/a>”—experienced mobile developers who, despite history with Xcode and a predilection for building native apps, are begrudgingly swapping out their tools in acknowledgement of the inarguable productivity benefits.<\/p>

Sure, AI-assisted developer tools can<\/em> be used to generate native iOS apps, but they’re not nearly as good at this as they are at generating e.g. React, whose developer experience advantage predates the LLM wave and has only since accelerated.<\/p>

[…]<\/p>

App Review has always long been a major source of developer frustration. Authoritarian yet inconsistent policy enforcement aside, it’s simply too hard to distribute software even to your own Apple devices<\/em>, let alone someone else’s. This isn’t new by any means, but as the time to build an app shrinks from weeks\/months to hours\/days, it feels more egregious—and thus like more of a liability—than ever before.<\/p><\/blockquote>\n\n

John Gruber<\/a>:<\/p>\n

Basically, the threat to Apple the App Store poses is NOT regulators coming for it. That’s a distraction. The threat, as I’ve always tried to argue, perhaps unsuccessfully, is that market forces will work against it eventually. The App Store should have been making developers (mostly) happy all along, not mostly miserable.<\/p>

And AI might be the disruption that brings about the “eventually”.<\/p><\/blockquote>\n\n

Kyle Hughes<\/a>:<\/p>\n

\n

The speed with which I abandoned a decade of manicured Xcode project & window management muscle memory in favor of a mess of VS Code clones taught me something about myself and my work and what is important.<\/p>\n<\/blockquote>\n\n

Jon Stokes<\/a>:<\/p>\n

\n

In normieland, where I still spend plenty of time both online and IRL, software’s newfound ability to write, draw, and speak like we humans is often taken as evidence that the machines are about to remake our entire society (again) and totally change the nature and value of labor (again). I don’t think the normies are wrong about this, but as flashy as Stable Diffusion<\/a> and ChatGPT<\/a> are, old heads know that the Robot Apocalypse has exactly one and only one horseman<\/a>: computer programs that can write computer programs<\/strong>.<\/p>\n

[…]<\/p>\n

If you were going to design a platform for the express purpose of teaching machines to code, it would probably be a cloud-hosted IDE plus execution environment that looks a lot like Replit. So while most of what you’ll read in this article and followups is applicable to all AI code generation tools more generally, I’ll be focusing on Replit’s toolset because right now it’s the richest and most advanced, and has the most potential for advancing the state-of-the-art.<\/p>\n<\/blockquote>\n\n

Previously:<\/p>\n