
Apple used WWDC 2026 to announce a lot of AI features that, frankly, you've seen before. Chatbots, text summarization, image generation — the kind of stuff already living on Android phones and inside Claude and ChatGPT. But according to Apple's best AI idea looks a lot like vibe coding, The Verge's David Pierce found something more interesting the moment he downloaded the first developer beta of iPadOS 26: an Apple Shortcuts experience powered by natural language that lets you describe what you want your phone to do and watch it get built in seconds.
Pierce describes typing "Send a text to Anna with three kissy emojis" into the Shortcuts app and seeing a working shortcut appear almost immediately. Simple, personal, genuinely useful. That's a different kind of AI pitch than "trust me, this changes everything."
The Details: What Apple Actually Announced
Apple Shortcuts has always been powerful and almost always been confusing. It's essentially a visual scripting tool that lets users automate tasks across apps. The ceiling is high. The learning curve has kept most people away. At WWDC, Apple's product marketing manager Cecilia Dantas called the AI-powered update "more approachable than ever" — and the idea is straightforward: describe what you want in plain English, and Apple Intelligence builds the shortcut for you.
The first developer beta is rough, though. Pierce tested several scenarios and ran into walls fast. A request for a shortcut that would activate Do Not Disturb when the Kindle app opened instead just triggered Do Not Disturb on tap. A more complex shortcut involving a timer with a "Stop Playing" action failed even after follow-up prompts. A photo-stitching shortcut understood the steps but couldn't execute them. Third-party apps pushed him back to the standard editor entirely, which points to a real dependency: developers need to build out App Intents support before any of this scales.
Safari got a related upgrade. The new OS versions let you create browser extensions the same way — describe what you want in natural language and the extension is generated. Pierce reports that simple tasks, like copying a page as a Markdown link, work well. The ceiling is lower than Shortcuts (no AI calls, no cross-app interaction in the current build), but the principle is identical: mold the device to fit how you actually work, without needing to build anything from scratch.
Pierce also notes that Apple is simultaneously using AI to automatically organize and categorize Safari tabs — a more passive feature sitting alongside the active extension-creation tool.
Why It Matters: The "Vibe Coding Your Phone" Angle
Here's the framing Pierce uses that stuck with me: these natural-language shortcuts are effectively vibe-coding projects. You describe intent, the system infers structure, and something functional gets built. It's the same loop developers use in tools like Cursor or GitHub Copilot — just pointed at personal automation instead of a codebase.
The irony Pierce points out is sharp. Apple has reportedly taken a hostile stance toward vibe-coding apps on its platform, and yet here it is shipping a feature that is, in essence, vibe coding for end users. The difference is that Apple controls the rails. It has unique access to location, app logins, and system-level permissions that third-party tools simply don't. If the feature works — and that "if" is still large — the depth of what's possible is hard for any standalone AI app to match.
The bigger strategic point is this: Apple already has what every AI company is desperately trying to build. A super app. A single platform that knows who you are, what apps you use, and how you move through your day. The iPhone is that platform. If Apple can make natural-language automation actually reliable, it doesn't need to reinvent the interface. It just needs to make the one you already have smarter.
That's a meaningful distinction from everything else being announced in the AI space right now. Most AI products are asking you to change your behavior. Apple is — at least in theory — trying to make your existing behavior easier.
What to Do Now
If you're an agency owner, marketer, or content strategist, this isn't a story you can safely ignore until the public release. Here's where to put your attention:
- Watch the developer beta cycle closely. The first iPadOS 26 beta is where Pierce tested this, and it's already showing the gap between concept and execution. The features that survive to public release — and how they change — will tell you a lot about what Apple is actually shipping versus what was a demo.
- Think about your own workflow automations. If natural-language Shortcuts eventually works reliably, a lot of the copy-paste, tab-switching, and cross-app grunt work your team does every day could be scripted by describing it out loud. Start mapping those repetitive tasks now so you're ready to automate fast.
- Pay attention to App Intents support. Pierce is clear that third-party apps are a blocker right now. If you build or recommend tools for clients, check whether those tools are investing in App Intents. This will separate the apps that play nicely with Apple's AI layer from the ones that get bypassed entirely.
- Don't overweight the Safari extension feature yet. Simple Markdown-copy-style extensions work in the first beta. Anything involving AI calls or cross-app behavior doesn't. Useful for light workflow tweaks, not a platform shift — at least not yet.
- Track how this affects search behavior. Safari is still a major traffic source for mobile. If AI-generated extensions make it easier for users to bypass page visits — grabbing content as Markdown, for example — that has downstream implications for how you think about content structure and AI search visibility strategy.
Background and Context: Where This Fits the Bigger Picture
We're two-plus years into the mainstream AI boom and most of what's shipping still looks like feature parity wrapped in new branding. Every major platform has a chatbot. Every productivity suite has a summarizer. The race to the bottom on "AI features" has been genuinely exhausting to watch from the conference circuit.
What Apple is attempting with Shortcuts is different in kind, not just degree. It's agentic AI applied to personal automation — the same category that Zapier, IFTTT, and a dozen well-funded startups have been chasing for years. The difference is platform depth. Apple doesn't need to ask for permissions the way a third-party automation tool does. It already has them. That's a structural advantage that no amount of venture funding replicates.
The challenge Pierce identifies is the same one every agentic AI system hits: developer adoption. Getting every app in the App Store to support Shortcuts thoroughly enough for this to be genuinely useful is, in his words, "an outrageously huge amount of work." Developers have real incentive to resist — if Shortcuts works perfectly, users never open their app. That tension isn't going away, and Apple will have to navigate it carefully.
I've watched agentic AI features get announced with enormous fanfare and then quietly recede into the background because the underlying infrastructure wasn't ready. This could follow that pattern. Or Apple's platform control could make it the exception. Either way, the automation layer of the iPhone is now officially in play — and that's worth tracking whether you're building apps, creating content, or just trying to stay ahead of where user behavior is going.
If you want a clean way to track how AI platform changes like this affect your organic visibility and content performance over time, Aergos gives you the reporting layer to spot shifts before they become problems.
Frequently Asked Questions
Related Articles
Glossary terms in this article
Brush up on the definitions.
A development style coined by Andrej Karpathy in early 2025 where the developer steers an AI coding assistant by intent and feedback rather than writing most of the code directly.
AI systems designed to act autonomously toward goals, using tools and multi-step reasoning rather than responding to single prompts.
Search experiences powered by large language models that generate conversational answers, synthesize information from multiple sources, and reduce reliance on traditional blue-link SERPs.

About Matt Weitzman
Senior SEO Strategist & Co-Founder
Matt has over 15 years of experience in technical SEO and digital marketing. He specializes in algorithmic recovery, enterprise architecture, and leveraging AI for content scaling. He is a frequent speaker at search marketing conferences.
More articles by Matt Weitzman

