Siri AI can now control third-party apps through Apple’s App Intents system, and Speech Central’s Siri AI integration is now live on iPhone, iPad, and Mac. After testing it with the iOS 27 Release Candidate, we learned that Siri’s third-party app integration works somewhat differently from what the earlier App Intents schemas suggested.
Speech Central now supports Siri AI playback through Apple’s .audio.play schema, can provide news through .audio.newsProvider, and can interpret approximate natural-language requests such as asking Siri to play an audiobook without specifying its exact title.
This is a follow-up to our earlier article about Speech Central’s early support for Siri AI and Apple Intelligence App Intents schemas. At the time, we were working from Apple’s developer documentation and the APIs available before the new Siri experience could be tested in practice. Now we can see what actually happens when Siri sends requests to a third-party app.
How Does Siri AI Work With Third-Party Apps?
Apple’s App Intents system gives Siri a standardized way to communicate with third-party apps. Instead of every app inventing its own vocabulary for Siri commands, apps implement schemas for common actions such as playing audio or providing news.
In simplified form, the process looks like this:
Siri AI → App Intents schema → information extracted from the user’s request → third-party app → app resolves and performs the action
One important thing we learned from real-world testing is that Siri does not necessarily resolve every natural-language request into a perfect, unambiguous item before handing control to the app.
The app may still receive an approximate description of what the user wants. It then needs to interpret that information and find the best match in its own library.
That distinction matters because it gives apps like Speech Central an important role in the intelligence of the overall interaction. Siri understands the request and routes it to the appropriate app and action, while Speech Central can use its knowledge of the user’s library to determine what should actually be played.
What Changed From the Earlier Siri AI Implementation?
Our initial implementation was based on the App Intents schemas Apple had documented before we could test the finished Siri AI behavior.
One of the most interesting discoveries from testing the iOS 27 Release Candidate was that Siri uses the more general .audio.play schema for playback rather than the .books.playAudiobook schema we had previously implemented.
This makes the integration more general than we initially expected. Instead of treating Speech Central only as an audiobook application, Siri can interact with it as an audio app capable of resolving and playing different types of content.
We adapted Speech Central accordingly, and the live implementation now supports the schema Siri actually uses.
Natural-Language Playback Is More Interesting Than Exact Titles
The most useful part of Siri AI integration is not simply being able to say an exact title. Traditional voice assistants could already handle commands built around precisely named items.
The more interesting case is when the request is incomplete.
For example, a user might ask Siri:
Play the audiobook about Alice in Speech Central.
The item on the Speech Central Shelf might have a much longer title, or the word “Alice” might appear only as part of its title or metadata.
Siri can recognize that the user wants Speech Central to play something and invoke the appropriate audio intent. Speech Central then receives the information Siri extracted from the request and searches its own content for the most appropriate match.
This division of responsibilities is important. Siri does not need complete knowledge of everything stored inside Speech Central. The app itself can perform the final matching using information and context that it understands better than the system does.
What Can Siri AI Do With Speech Central?
With the current integration, Siri AI can use Speech Central for several types of interaction:
- Natural-language playback: Siri can invoke Speech Central through the
.audio.playschema, while Speech Central resolves the requested content against items on the user’s Shelf. - Approximate content matching: requests do not always need to contain the exact title of an item. Speech Central can interpret the information supplied by Siri and find the closest relevant content.
- News and headlines: Speech Central supports the
.audio.newsProviderschema, allowing Siri to use the app as a source for spoken news content. - Audio preferences: support for
.audio.updateAudioAffinityextends the integration to preferences such as favorites.
These schemas also make the integration less dependent on users memorizing a specific set of commands. The goal is for users to describe what they want naturally and let Siri and Speech Central determine how to perform it.
Siri AI and Speech Central as a News Provider
The .audio.newsProvider integration is another interesting part of the new system.
Speech Central is designed not only for books and documents but also for consuming web content and news through text-to-speech. That makes news intents a natural fit for the app.
Through the standardized App Intents schema, Siri can recognize Speech Central as an application capable of providing news and route an appropriate request to it.
This is also an example of why Apple’s schema-based approach is significant for third-party developers. Siri does not need custom knowledge of every feature in every application. Apps declare support for standardized capabilities, and Siri can connect a user’s natural-language request with those capabilities.
What We Learned From Testing Siri AI in a Real App
The biggest lesson from implementing and then testing Siri AI support is that App Intents are not simply a collection of predefined voice commands.
There are several layers involved. Siri interprets natural language, determines which app and capability should handle the request, and supplies structured information through an App Intent. The receiving app may then need to perform its own interpretation and matching before executing the action.
For developers, that means implementing the correct schema is only part of the job. A good Siri AI integration also needs to handle imperfect or incomplete input gracefully.
In Speech Central’s case, this is particularly relevant because a user may have hundreds of books, articles, documents, feeds, and other items on the Shelf. The phrase a person naturally uses when speaking to Siri will not necessarily correspond exactly to the stored title.
Allowing Speech Central to perform that final resolution makes natural-language interaction considerably more useful.
Why App Intents Matter for Siri AI
The broader significance of App Intents is that they provide a bridge between Siri’s language understanding and the specialized capabilities of third-party apps.
Siri can understand what a person is trying to accomplish without needing to understand every detail of an app’s internal data. At the same time, the app does not need to build its own complete voice-assistant system.
Each side handles the part it knows best: Siri interprets the user’s language and selects an action, while the app applies its domain knowledge and performs that action.
Our experience with Speech Central also shows why testing against the actual Siri implementation matters. The schemas and behavior used in practice can evolve, and the most effective integration may differ from what developers initially expect from the documentation alone.
Siri AI Support Is Available in Speech Central
Speech Central’s Siri AI support is now live, including support for Siri-driven audio playback and news integration through Apple’s App Intents system.
We expect these integrations to become more capable as Apple expands Siri AI and its standardized App Intents schemas. We will continue adapting Speech Central as those capabilities evolve.
If you are interested in the implementation history and the schemas we originally prepared for, read our earlier article: Speech Central Is Ready for Siri AI with Apple Intelligence App Schemas.