Vibe Coding in the Apple Ecosystem is Everything You Need to Know About AI ⭐️

2026 will be remembered for all kinds of reasons. Key among them is that vibe coding became real this year. It is astonishing how quickly this technology has evolved and how accessible it’s become in such a short time.

Less astonishing is the continued pushback from older, more well-established technical people who still cling to outdated beliefs about AI. In this sphere, the chief concern is that while vibe coding may be fun for small, personal projects, it will never lead to supported, maintainable code bases for more professional apps that will be used by millions out in the world or in businesses.

That’s cute. But in the blink-and-you’ll-miss-it world of AI, things move much more quickly than ever before. In the 1990s, upstart Netscape was going to give creaking, ancient Microsoft the heave-ho by working on “Internet time” and shipping major software products each week, not once every two to three years. But today, that speed is slow, not fast, even for traditional software. Many of the apps I use are updated at least a few times each week, if not more.

AI takes this to another level, and it’s not an exaggeration to say that it’s improving exponentially faster than any technology before it. You can see this most clearly when you compare it to the non-AI developer advances that Apple and Google both outlined at their respective developer shows alongside whatever AI advances. For example, they are both promoting apps with dynamic layouts that don’t target specific devices that will adapt on the fly to whatever form factor they’re run on. Exciting, right? Sure, but it was just as exciting when Microsoft announced this same capability in Avalon, the presentation technology in Longhorn, in October 2023. It’s like the industry has been treading water for over two decades.

Which would be true if it weren’t for AI. With AI, increasingly, anything you can imagine making is suddenly possible. And if it isn’t, just blink, because that capability is coming soon. The time frame is measured in days or weeks, not months or years. We’re already at the point where using AI to create things is only controversial for those people who were already making things in more tedious and time consuming ways and expecting to be paid for their ADHD-like focus on some individual skill that AI just turned into a commodity. But for the rest of us, this is magical and life changing. And it just keeps getting better.

Even experts in whatever fields, understandably worried about their futures and careers, should acknowledge that AI giving their skills away is a net gain for all of us. As a writer, I am not outraged that AI is being used to create words, nor will I be surprised when AI inevitably wins awards for that writing. Moving from charcoal and paint on a cave wall to writing instruments and paper, mechanical typewriters, dedicated word processors, word processing software on computers, note-taking apps on phones, and whatever else didn’t happen overnight, but each step in this evolution expanded mankind’s ability to create and share ideas through words. As it does everywhere else, AI is expanding that ability more than ever before, and faster than ever before.

This can be writ large, and it can be writ small. I know fully grown adults of my generation who have whatever skills but can’t put together a grammatically correct sentence. I’m sure the theory when they grew up was that this would never matter because they weren’t going to be writers anyway. (I had this same theory about algebra early on, which I think about every quarter when I report on multiple financial reports and have to do that math.) But everyone writes. Decades ago, it could have been through postcards and letters, or whatever. Today, it might be something as simple as a text message or a social media post. Not being able to write well is almost a handicap, something that will make you seem smaller, somehow, or less intelligent. It will impact how people see you. So having an on-device function to correct grammar and spelling on the fly, no matter where you’re entering those words, isn’t a nicety anymore. It’s a requirement, and for all of us. Even the professional writers.

But that’s AI assistance writ small. Bigger, more impactful uses of this technology might include the creation of new works, plus the editing of existing works to meet length, tone, or other requirements. But it might also include the creation of tools you can use to make those words. Which is what I’ve been toying around with.

Over the past several years, I’ve written many versions of my .NETpad and WinUIpad apps, which began as Notepad clones, in various programming languages and app frameworks. I toiled over this code, as programmers do, and some versions of these apps took many months to create. There were many highs, but also far too many lows.

Over the past few months, I used AI to complete the latest version of those apps, and, more impressively, I used AI to create wholly new versions of those apps, and more complicated Markdown editors, out of thin air. Not really, as grumbly old-timers will keep pointing out, because that AI had to be trained on something, and that something was made by people, many people, over long periods of time, and … Yes, yes, Grandpa. We know. You used to walk miles through the snow every day to get to school too. We get it.

Suddenly, this capability is astonishing.

It wasn’t always this way. It went from fascinating and frustrating in late 2025 to dramatically more interesting and less frustrating in early 2026 to just being magical by mid-2026. And it’s happening everywhere. Google, Microsoft, and Apple just held their respective developer shows back-to-back-to-back in May and early June, and each of them announced and/or released impressive vibe coding advances. Google AI Studio, for example, can vibe code web apps and, now, native Android apps. Microsoft’s Windows Developer Skills can create modern WinUI apps. And now Apple is doing the same for native SwiftUI apps across its platforms using Xcode.

Those are for professional developers for the most part, but they are good enough that even non-developer enthusiasts can use each of them now to, yes, create their own apps. In the baby steps department–writ small, I guess–these companies are also easing non-technical consumers into vibe coding, even though none of them are using that term, most likely worried about scaring anyone. Google will let anyone vibe code custom widgets in Android 17, for example. And Apple will let anyone vibe code custom extensions for its Safari web browser. It’s happening, and not just for developers. For everyone.

Of course, we still have to get from here to there, from the frustrating past to the it-just-works future. It’s coming together quickly, as noted, but it’s also happening right now as I write this and you read it. We’re experiencing it in real time. And rather than fear these incursions into my expertise (writing) and interests (programming), I choose to embrace them whether I need them for my own work or not. And I do that because I know this will be transformative for those who do need it. I don’t need AI for writing words, and I won’t use it that way, but I do need it for grammar and spell checking. And I don’t need AI to create apps, though I will absolutely use AI for that purpose because it is dramatically faster and will lead to higher quality in less time.

I wrote several articles recently, linked above, about my experiences vibe coding over the past month or so. Last night, after watching WWDC26: Platforms State of the Union–the real WWDC 2026 keynote, the one for developers–twice, I started down two paths. The first was to write about how the Apple and Google developer initiatives these days compare to Microsoft’s approach when it was the most influential company in this market; that one is still in progress. And the second was to use the Xcode 27 beta Apple just announced to vibe code an app. I figured this would require an article of its own, similar to those previous articles. But it does not. In part because it just works.

This is what I did.

After installing the Xcode 27 beta and linking it to Anthropic Claude Code, I started a new, blank project. This is a new Xcode capability in that you just get into the editor without specifying anything related to the project, like its name, storage location, and target platforms. Then, I opened the Coding Assistant pane, which is basically GitHub Copilot but for Xcode, and I wrote a simple prompt:

I would like to create a Markdown editor that looks and works much like iA Writer and Typora. It should be simple and minimalistic, offer a way to export to HTML and PDF, and have a WYSIWYG editing mode by default.

It was 5:43 pm.

The AI told me it would help. It could see that I was using a new, blank Xcode project with no platform target, and it examined the macOS and iOS 27 APIs to ensure that they contained the capabilities it needed. It stepped through a plan, which it created, and presented me with three options tied to editing style (hide Markdown syntax markers by default for a WYSIWYG view), document model (a proper file-based app), and view notes (live editor with no separate preview view). 10 minutes had gone by. And then it was ready to get to work.

I, meanwhile, went to dinner with my wife. When I got back, I woke up my MacBook Air and saw that it had finished making the apps hours earlier. Based on the timestamps in the chat, it took less than 6 minutes.

The app worked the first time Xcode launched it. It even works well, with the WYSIWYG view and HTML and PDF exporting features I requested, plus multiple document/tab support and many other things one just gets from SwiftUI/macOS. It’s a full-featured, fully functional Markdown editor. I guess I could adapt it in different ways, have the project target the iPad too, and whatever else. But the point here is that this is analogous to the vibe coding capabilities I see elsewhere. Part of it is that I used the same underlying AI for many of these projects (though I, of course, used Gemini for the Google-based work). But more important, I think, is that Apple, like Microsoft and Google, grounds these tools in its specific documentation and knowledge base so that they just do the right thing and don’t get distracted by a web full of contradictory and non-pertinent information. Nine months ago, I struggled to get AI to stay on track with this kind of project. Today, it just coughs out a real app in minutes.

Internet time. How quaint.

This came up earlier for obvious reasons, but Arthur C. Clarke was correct when he noted that any sufficiently advanced technology is indistinguishable from magic. That’s what this is, magic, plain and simple.