A month before WWDC 2026, the most interesting story in iOS isn't a hardware leak — it's that Apple Intelligence is about to stop being a one-vendor system.
Multiple reports this week confirmed what's been circling for months: iOS 27 will introduce an Extensions framework that lets users pick the model behind Siri, Writing Tools, and Image Playground. ChatGPT was the only option in iOS 26. With iOS 27, Anthropic's Claude and Google's Gemini are both being internally tested as drop-in replacements, and the picker will live in Settings the same way default browsers and mail clients do today.
Why this matters more than the headline#
A model picker in Settings sounds like a user feature. It's actually a developer feature dressed up as one.
If your app does anything with on-device intelligence — summaries, drafting, classification — you've been coding around the assumption that "the model" is whatever ChatGPT-flavored thing Apple ships. Some of you wrote prompts that work great with GPT and break weirdly when handed to a different family. Some skipped the integration because you didn't want vendor lock-in.
iOS 27 Extensions changes both calculations. Your prompts have to survive being routed to whatever model your user picked. And the integration cost drops because you're not committing to a vendor — you're committing to Apple's Extensions surface.
The 30-day window#
The keynote is Monday, June 8. The first developer beta drops the same week. From past cycles, the rough timeline looks like this:
Extensions API documented. Sample code shipped. Read the WWDC schedule the day it drops.
Indie apps add Extensions support and ship to TestFlight. Editorial team starts watching.
Release candidates. Submission queue gets crowded. Reviewer tolerance for novel APIs drops.
iOS 27 ships. Apps already on the picker get the launch-day bump.
Apps that ship Extensions support in the first wave get a meaningful discoverability bump — Apple has always rewarded early adopters of new frameworks with editorial features, and a "works with your AI of choice" angle is exactly the kind of pitch the App Store editorial team picks up.
Apps that wait until September are competing in a crowded "added Apple Intelligence" wave with a hundred other writing apps and journals.
What the Stora pipeline does for this#
You cannot ship a fast iteration loop on Apple Intelligence without a fast submission loop. The bottleneck during a beta cycle is rarely the code — it's everything around the code. Updated screenshots for the Extensions UI. A reworked store listing that mentions the new feature without tripping the AI consent rule. A privacy manifest reflecting which providers your app can route to. A build that survives the first reviewer who's never seen Extensions before.
Stora was built for this loop. The screenshot generator regenerates your localized set when the UI changes. The AI store listing rewrites your description against current iOS metadata rules. The compliance engine flags the AI consent disclosure before review does. It gives back the days you'd otherwise burn on submission churn — and during a beta cycle, those days are the game.
The practical takeaway#
If you're shipping anything that touches Apple Intelligence in iOS 27, start the work now. Don't wait for the keynote. Read the WWDC 2026 session schedule the moment it drops on June 6, mark the Extensions session, watch it the day it airs, and have a TestFlight build by the end of June.
The model picker is going to be the headline. The apps that win are the ones already on the picker by launch.