AI, in your work.

What the preview opens up

On November 3, Apple announced a Developer Preview for integrating apps with Siri AI through App Intents in the 27.3 betas of iOS, iPadOS, macOS, and visionOS. This access lets developers prepare app content and actions for the assistant to use. Apple's announcement.

The company also introduces skills to guide the combination of several actions and extends access to entities in Spotlight's semantic index beyond existing schemas. These are capabilities described by Apple; this article provides no independent evaluation of their performance.

Availability in Spain depends on the platform

The Siri AI guide, updated on November 5, confirms this distinction within the European Union. Spanish is supported; on available platforms, the device and Siri languages must match. Siri AI requirements and availability.

Compatible hardware and early access

Hardware compatibility is another requirement. Apple's Intelligence compatibility list includes, among others, Macs with M1 or later, iPhone 15 Pro and Pro Max, iPhone 16 models or later, and iPads with M1 or later. Owning one of these devices does not remove Siri AI's regional restrictions. Compatible devices.

The two platform lists should be distinguished: watchOS appears in the general Siri AI beta, but not among the systems announced for this integration preview. Siri AI is also unavailable to users under 13. Siri AI guide.

Apple expects the new integrations to reach customers in the coming months and App Store submissions to open in spring, without specifying a day. It also promises further details about terms and usage policies. This indicative timetable does not establish a confirmed date for a business deployment. Announced timetable.

Limits that affect a future evaluation

Support documentation updated in November warns of variable daily limits for features using server-side models, including Siri AI. It specifies no universal allowance and announces expanded paid access for the future. Limits described by Apple.

The terms allow limits to change and a feature to be temporarily restricted when a limit is reached. An evaluation should therefore record requests that are limited or incomplete alongside successful responses. This last point is Nexo's methodological recommendation. Usage terms.

Nexo's interpretation: define actions before measuring them

For a workplace application, our interpretation is that the first step is to define what each action can read or change. Finding a record, preparing a draft, and submitting a request have different consequences. A future integration should be evaluated against explicit criteria for permissions, ambiguity, and confirmations.

A convincing conversation would not, by itself, provide sufficient evidence to accept an operation. The proposed evaluation would need to check which record was selected, what changed, and whether the change matched the user's intention and permissions.

Hypothetical example: an internal request app

A fictional application makes those criteria concrete. The responses in this table are requirements proposed by Nexo for a future evaluation; they do not describe demonstrated capabilities of Siri AI or an existing business product.