Siri AI Developer Preview: What Apps Can Do With App Intents
The Siri AI developer preview lets app developers test how Siri can find app content and carry out actions through App Intents. Apple introduced the preview on November 3 alongside the iOS 27.3, iPadOS 27.3, macOS 27.3 and visionOS 27.3 betas. The preview gives teams time to try an integration before Apple opens App Store submissions for these experiences in the spring.
Apple’s developer announcement distinguishes building and testing an integration from releasing it to customers. If you own an iPhone and are waiting to use these capabilities in a favourite app, the app developer still has to implement and ship the integration. Seeing the beta version number alone tells you little about support in a specific app.
What the Siri AI developer preview adds
App Intents expose named actions and app content to system features. Think of a parcel tracker. The app might expose an action that retrieves a shipment’s latest status and an item representing a particular parcel. A user could ask about the parcel in ordinary language, and Siri could call the action the developer provided. The app still controls the logic for finding a shipment and checking who can access it.
Apple says Siri AI can now surface entities from the Spotlight semantic index beyond the types covered by existing schemas. It also introduced skills for combining several actions into an experience. In a to-do app, for example, a useful starting point would be finding a list by name and adding an item to it. A combination could help with a longer request that uses those two actions together. These examples illustrate the model and would depend on what the app implements.
Which devices and languages can test it?
Apple says the Siri AI beta requires an Apple Intelligence-enabled device set to a supported language. Its announcement lists English, French, Japanese, Korean, Portuguese and Spanish among the available languages and says some features vary by region or language. The company also lists watchOS 27.3 among beta platforms for Siri AI. Check the requirements for the specific platform and capability you’re testing before scheduling work across a team.
For an app used in India, one practical test is to use natural requests in supported English and vary how people say names or refer to an item. Test an ambiguous name as well as the ideal example. If a user asks for the status of a delivery and has two parcels with similar labels, make sure the experience can recover cleanly when it needs clarification.
Keep a record of the beta build, device and language for each test. A failure on an unsupported configuration is a different problem from a request that reaches your app with the wrong item. That small bit of context can save a second developer from repeating the same investigation.
A good first test for an app team
Choose one action with a visible result and few side effects. Checking an order status is an easier place to start than placing a paid order. Try it on an eligible device with the right beta, then check what happens if the app is signed out, the requested item has been deleted or a network request fails. Make sure the action respects the same access controls the user encounters inside the app.
When testing a change that writes data, check that it happens once, that the result is clear and that the user has an obvious way to correct it. Apple says more guidance on terms and usage policies will follow. Use the preview to evaluate whether the action is worth exposing and track the rollout in Apple’s developer updates. Bootpoot’s Tech News coverage follows platform and device changes.
Illustrative photo by Jayanth Muppaneni on Unsplash.
