Siri AI and App Intents: make your app discoverable in conversation

Discovery no longer starts only in the App Store

Apple opened the Siri AI Developer Preview to all developers on November 3, 2026. Apps can expose their content and actions through App Intents. A person should be able to ask about a delivery, book a reservation, or place an order in ordinary language without navigating a menu first.

This is more than a new voice control. It changes the first contact with an app. App discoverability has traditionally started with an App Store search, an ad, or a link. It can now begin in the middle of a conversation. Siri AI understands an intention and looks for an app whose content and actions fit the request.

That creates a new job for product teams. An app must do more than look good and present the right App Store metadata. Its product model has to become understandable to the operating system.

The Developer Preview is an opportunity, not a launch signal

Siri AI is available in the iOS 27.3, iPadOS 27.3, macOS 27.3, watchOS 27.3, and visionOS 27.3 betas. Apple says customer availability is a few months away and plans to open App Store submissions in spring. Detailed terms and usage policies will follow later.

One limitation matters especially for European businesses. Siri AI is currently unavailable on iPhone, iPad, and Apple Watch in the European Union. A production launch for the EU market cannot be promised responsibly today.

This is still the right time to prepare. App Intents do not serve Siri alone. They connect an app with Spotlight, Shortcuts, widgets, controls, and other system surfaces. Clear structure creates value before Siri AI reaches every market.

App Intents are the product inventory of an app

Every useful app contains verbs and things. A delivery app can find an order and show its status. A property app can search for a listing, save it, and arrange a viewing. A service app can report a fault and show the next available appointment.

App Intents describe the verbs. App Entities describe the things. App Enums give fixed choices a defined meaning. During compilation, the framework generates information that Apple Intelligence, Siri, and other system features can discover and use.

The quality of this structure affects the quality of the answer. An intent called “Perform action” helps neither a person nor a language model. A clearly named “Check delivery status” intent has an understandable goal, defined parameters, and an expected result.

Treating App Intents as a technical add on wastes their potential. They are a machine readable description of what the product can actually do.

App Entities make content findable

An action is rarely enough. Siri needs to know which project, ticket, reservation, or order a person means. The app models those business objects as App Entities.

A good entity has a stable identity and the properties required for selection and display. A reservation may need a location, time, and status. A project may need a name, client, and last activity date.

This is data modeling, because conversational interfaces are not reliable when the underlying product model is vague. Work on Siri AI therefore does not begin with sample phrases. It begins by identifying the objects customers understand and the relationships that matter to them.

The Spotlight index becomes a discovery channel

Apple Intelligence uses Spotlight semantic search to find app content even when a person does not know its exact name. IndexedEntity can place local content in that semantic index. Siri can then search by meaning instead of matching strings alone.

This matters for app marketing. Traditional App Store Optimization works with the title, subtitle, keywords, screenshots, and reviews. Semantic app discovery adds actual product content to that layer. A travel app is not visible only for the word travel. It may find a saved route described as “our weekend by the coast.”

The index needs maintenance. New entities must be added, changed content updated, and deleted content removed. Stale information is especially harmful in conversation because a fluent answer can easily sound authoritative.

When data is too large, changes too quickly, or lives only on a server, the app can answer its own search queries. Teams need to decide which content belongs in a local semantic index and which search should remain under their control.

App Schemas translate product logic into everyday language

Schemas define familiar kinds of actions and content. They create a contract between the app and the system. Apple Intelligence uses that contract to understand what an intent does and how spoken language maps to it.

This standardization matters because customers do not speak in backlog terminology. They do not say “Run intent number 14 with parameter B.” They say “Book the same table as last Friday” or “Show the invoice from my latest job.”

An app should adopt the relevant schemas instead of inventing every meaning from scratch. The mapping still has to be honest. A schema that does not fit the product will not improve discoverability. It creates false expectations and more follow up questions.

Onscreen context closes the last gap

Many requests refer to what is visible right now. “Share this photo,” “book the second appointment,” or “open the invoice here” only works when Siri can connect the displayed content to App Entities.

Apple provides links between views, user activities, and entities for this purpose. The screen remains private app content, but it receives machine readable meaning. Siri can resolve a phrase like “this item” to the correct object.

That makes design and development one task. An element that looks unambiguous is not automatically unambiguous to the system. Strong accessibility, clear states, and accurate entity associations become parts of the same user experience.

An executable answer needs trust

The main difference between AI search and an App Intent is the consequence. Search returns information. An intent may change, send, book, or purchase something.

Every action therefore needs a deliberate boundary. A status request can answer directly. An order, public share, or change to collaborative content needs a clear confirmation. The experience also has to show which app is acting and which object is affected.

A good Siri integration does not minimize every follow up question at any cost. It optimizes for a fast and safe decision. Trust grows when the result, owner, and consequence remain clear.

Conversational discoverability can be measured

Siri AI does not offer a public ranking report comparable to App Store keyword positions today. Teams can still work with evidence. They can define common jobs and test whether Siri finds the right entity, selects the right intent, and completes the task without unnecessary correction.

Useful signals include completion rate, number of follow up questions, time to result, and common failure reasons. Coverage matters too. Which valuable customer jobs are structured, and which remain hidden behind visual navigation?

These tests belong in the product process. An integration that works only with one memorized sentence is not a natural language interface. It is a hidden voice command.

The first sprint needs one real customer job

A practical start does not require mapping the entire app. Choose one frequent task with a clear result. Model the entities involved, build one intent, apply a fitting schema, and keep the search data current. Then test different wording, incomplete details, and situations with similarly named objects.

Consider a field service app. The first customer job could be checking the status of a work order. Siri does not need a complete ERP integration for that. It needs a uniquely findable work order entity, a readable status, secure authorization, and a clear response.

Once that path works reliably, add the next job. This creates a conversational experience from real product capabilities instead of an impressive demo that nobody uses in everyday work.

Our conclusion

Siri AI turns App Intents into a new surface for mobile apps. App Entities, App Schemas, Spotlight semantic search, and onscreen context determine whether a product can be found and used safely through natural language.

For app development and app marketing, this is the same job seen from two directions. Marketing wants to be visible at the right moment. Development has to structure the product so the system understands that moment.

A team that clarifies its product model now is not building only for a voice assistant. It is building an app whose content and actions remain understandable in the next generation of AI search and conversational interfaces.

Primary sources

Apple Developer News: Siri AI is now open to all developers describes the Developer Preview, supported platforms, regional availability, and planned timeline.

Apple Developer Documentation: Apple Intelligence and Siri AI documents App Entities, Spotlight indexing, schemas, onscreen context, transfers, and donations.

Apple Developer Documentation: App Intents explains how actions and content become available to Apple Intelligence, Siri, Spotlight, Shortcuts, and widgets.

WWDC26: Build intelligent Siri experiences with App Schemas demonstrates entity resolution, IndexedEntity, App Schemas, and natural language testing.