On November 3, 2026, Apple opened a Siri AI developer preview that lets any developer connect an app's content and actions to Siri AI using App Intents (Apple Developer News). It's available in the iOS 27.3, iPadOS 27.3, macOS 27.3 and visionOS 27.3 betas.

The preview is early, though, and it has real limits. This post covers what you can try now, how the pieces fit together, and what to watch out for.

The catch first

Before you plan anything, know these constraints:

  • It's beta-only. App Store builds can't use the preview, and TestFlight testers don't get the preview features, even if your app already uses App Intents.
  • Shipping starts in spring. App Store submissions for apps using the new integrations open then.
  • Hardware and language requirements apply. Siri AI needs an Apple Intelligence-enabled device set to a supported language.

So this is the time to experiment and design, not to ship.

What's new

Through App Intents, your app can expose content and actions to Siri AI. Apple highlights three capabilities:

  • Personal context understanding
  • On-screen awareness
  • Natural-language requests instead of fixed commands

Apple also introduced app skills, which guide Siri through several actions when a request is more complex than a single intent.

Xcode includes a Siri Integration Advisor that suggests improvements to your App Intents. Start there on an existing app.

The building blocks

If you've used App Intents for Shortcuts or Spotlight, the model is the same:

  • AppIntent: an action users can perform
  • AppEntity: a structured piece of your app's content
  • EntityQuery: how the system finds and resolves entities

Here's a minimal example for an order-tracking app. OrderStore stands in for your own data layer.

import AppIntents struct OrderEntity: AppEntity { static let typeDisplayRepresentation: TypeDisplayRepresentation = "Order" static let defaultQuery = OrderQuery() let id: UUID let status: String var displayRepresentation: DisplayRepresentation { DisplayRepresentation( title: "Order \(id.uuidString.prefix(8))", subtitle: "\(status)" ) } } struct OrderQuery: EntityQuery { func entities(for identifiers: [UUID]) async throws -> [OrderEntity] { try await OrderStore.shared.orders(withIDs: identifiers) } func suggestedEntities() async throws -> [OrderEntity] { try await OrderStore.shared.recentOrders() } } struct CheckOrderStatusIntent: AppIntent { static let title: LocalizedStringResource = "Check Order Status" static let description = IntentDescription("Looks up the status of an order.") @Parameter(title: "Order") var order: OrderEntity func perform() async throws -> some IntentResult & ProvidesDialog { .result(dialog: "Your order is \(order.status).") } }

A request like "Where's my latest order?" can now resolve to an entity from suggestedEntities() and run the intent, without the user opening the app.

Design tips

1. Expose a few good actions, not everything. Pick the tasks users do most often. Give each intent clear parameters and a predictable result.

2. Model your data properly. Siri can only reason about what you describe. Well-defined entities and display representations make matching far more reliable.

3. Gate sensitive actions. Payments, account changes and personal data shouldn't run on a voice request alone. App Intents let you require authentication and ask for confirmation:

struct CancelOrderIntent: AppIntent { static let title: LocalizedStringResource = "Cancel Order" static var authenticationPolicy: IntentAuthenticationPolicy = .requiresAuthentication @Parameter(title: "Order") var order: OrderEntity func perform() async throws -> some IntentResult { try await requestConfirmation() try await OrderStore.shared.cancel(order.id) return .result() } }

4. Plan for ambiguity. Natural language is messy. Handle missing parameters, multiple matches ("which order?") and unexpected input gracefully.

5. Keep behavior consistent. An action should work the same whether it's triggered from your UI or from Siri. Share the underlying logic rather than duplicating it.

6. Expose only what's necessary. Respect existing access controls, and don't surface data just because you can.

Try it today

  1. Install the 27.3 beta on a supported device.
  2. Run Xcode's Siri Integration Advisor against your existing intents.
  3. Start with one read-only intent, like the order lookup above, before adding actions with side effects.
  4. Read Apple's documentation on app skills for multi-step flows.

Closing thoughts

The shift here is that your app becomes a set of capabilities Siri can call, in addition to a place users navigate. How much that matters depends on adoption once the spring submissions open. In the meantime, clean App Intents and well-modeled entities are cheap to build and useful for Shortcuts and Spotlight anyway.

If you're planning a new product, it's worth thinking about this earlier in the design process rather than retrofitting it later. Working with an experienced mobile app development company can help you decide which actions to expose, structure your data, and build an architecture that's ready for features like this as they mature.

Are you planning to adopt Siri AI integration? Let me know in the comments.