WWDC 2026: iOS 27, the New Siri AI, and What It Means for App Developers

Apple opened WWDC 2026 with the announcement most developers had been waiting two years for: a genuinely rebuilt Siri, now branded Siri AI, sitting on top of an expanded Apple Intelligence layer. The 9to5Mac Happy Hour crew, Benjamin Mayo and Chance Miller, ran through their first impressions from Apple Park, but the part that matters for anyone shipping iOS and Android apps is less about the keynote demos and more about what shows up in the developer betas this week.

Here is what changed, what it means if you maintain apps on both platforms, and how to start planning your migration.

The platform versions

This year's release train is iOS 27, iPadOS 27, and macOS Golden Gate, with the usual companion bumps to watchOS, tvOS, and visionOS. The version numbering stays aligned across iPhone and iPad, which Apple settled into a couple of cycles ago, so you can keep treating them as a single deployment target with size-class differences rather than separate OS generations.

The first developer betas are seeded now, with public betas typically following in July and a general release in the fall. If your release cadence mirrors most teams, that gives you roughly a summer to test against the new SDK before users start updating en masse. The practical advice has not changed: install the beta on a secondary device, not your daily driver, and pin your CI to a known Xcode version rather than auto-adopting the beta toolchain, because the beta Xcode will happily rewrite project settings you did not intend to change.

Siri AI and Apple Intelligence for developers

The headline is Siri AI, which Apple describes as a profoundly more capable and personal assistant. The 2025 version of this promise slipped repeatedly, so the relevant question for developers is whether the new assistant exposes anything you can actually build against, or whether it stays a system feature you can only feed through existing entry points.

The path Apple has been laying down runs through App Intents. If you have not adopted the App Intents framework yet, this is the year it stops being optional. App Intents is how you describe your app's actions and entities to the system in a structured way, and it is the surface Siri AI uses to understand what your app can do, pull content into its personal context, and chain actions across apps. An app that has modeled its core actions as intents gets to participate in the assistant's on-screen awareness and multi-step requests. An app that has not is effectively invisible to it.

For cross-platform teams, this is the familiar split. On Android you have been wiring up App Actions and shortcuts.xml, intent filters, and increasingly the Gemini-driven app integration surfaces. The mental model transfers reasonably well: you are describing capabilities and entities to a system assistant rather than building the assistant yourself. The work does not, however, share code. You maintain an App Intents layer in Swift on the Apple side and the equivalent capability declarations on the Android side, and you keep them semantically in sync so a user gets the same set of actions regardless of which phone is in their hand.

Apple Intelligence itself is positioned as bringing AI capabilities into everyday experiences, which in developer terms means the on-device foundation models, Writing Tools, Image Playground, and the Genmoji and semantic-search APIs continue to expand. If you integrated the foundation model APIs last year, expect new capabilities and, as usual, some deprecations to read carefully in the release notes. The on-device model approach keeps inference local, which is the trade-off worth understanding: lower latency and no per-request server cost against a hard ceiling on model size and capability compared to a cloud call. For features where that ceiling is too low, Apple's Private Cloud Compute remains the escape hatch, and the design pattern is to let the system decide on-device versus cloud rather than you hardcoding it.

The EU asterisk you have to plan around

The most consequential line for many teams is buried in the announcements: due to the Digital Markets Act, Siri AI is delayed in the EU for iOS 27 and iPadOS 27. This is the same fragmentation pattern that hit Apple Intelligence features in earlier releases, and it is now a permanent part of how you have to think about feature availability.

Concretely, this means you cannot assume a feature is present just because the OS version supports it. Region becomes a capability check. If you build a flow that depends on Siri AI handling an App Intent, you need a graceful fallback for users in EU member states, and you need to test that fallback, because a meaningful slice of your install base will hit it on day one. The cleanest approach is to treat assistant integration as progressive enhancement: your app must remain fully usable when the assistant capability is absent, with the deep integration layered on top only where the system reports it available. Do not gate core functionality behind a feature that ships in some regions and not others.

This also has product and support implications beyond code. Your release notes, your help documentation, and your support team all need to know that an advertised feature legitimately does not exist for EU users yet, so you are not chasing phantom bug reports.

Child safety and services

Apple also previewed new child safety features and a round of updates across its services with new intelligence experiences. For most third-party developers these are less of a direct integration concern, but if your app handles user-generated content, age-gated features, or family sharing, it is worth reading the child safety documentation closely once it posts. Apple tends to ship new APIs and entitlements alongside these initiatives, and the App Review guidelines usually get updated language to match, which can affect what you are required to implement.

How to spend your summer

The migration checklist this cycle looks like this. Pull the iOS 27 and macOS Golden Gate SDKs and do a clean build to surface deprecation warnings early. Audit your App Intents coverage and fill the gaps, because that is the gateway to everything Siri AI touches. Add region-aware capability checks so EU users get a coherent experience without the delayed assistant features. Re-test any existing Apple Intelligence integrations against the new foundation model APIs. And as always, watch the WWDC session videos for the frameworks you actually depend on rather than trying to absorb the entire firehose.

The broader pattern holds steady. Apple is steadily turning the assistant from a voice feature into a structured integration surface, and the apps that model their functionality cleanly as intents and entities are the ones that get to ride along. The fragmentation tax from regulation is now a fixed cost of doing business, not a temporary inconvenience, so building region awareness into your feature flags from the start saves you from bolting it on later. You can read Apple's full set of announcements on the Apple Newsroom and catch the full first-impressions discussion on the 9to5Mac Happy Hour podcast.