SiriKit Is Deprecated. Your Booking Software Has a New AI Access Problem.

Apple's iOS 27 public beta launched July 13, 2026 -- one day ago. Buried in the developer notes is a confirmation that changes the calculus for every local service business currently using booking software to capture customer appointments.

SiriKit is officially deprecated. App Intents is now the only framework for Siri third-party integrations. Not the preferred path, not the better option -- the only one. The 2-3 year support window means SiriKit integrations don't vanish immediately, but the legacy fallback period is fixed. When it ends, booking software that hasn't migrated loses Siri AI access with no patch.

Until yesterday, "migrate to App Intents" was on vendors' roadmaps with no forcing function. Today it has a deadline.

What SiriKit and App Intents actually handle differently

SiriKit handled voice-command integrations: "Hey Siri, call my plumber," "Add an appointment to my calendar." These were explicit, single-step commands where Siri relayed information to the app and stopped.

App Intents handles AI-native workflows. When Siri AI -- the Gemini-powered, camera-aware assistant Apple announced at WWDC 2026 -- processes a request like "book me an HVAC appointment for Thursday afternoon," it doesn't relay a command and stop. It navigates your booking software, selects an available slot, confirms the appointment, and returns a summary. The entire transaction completes without a human touching a browser.

That navigation requires App Intents. SiriKit doesn't have the interface for it. A booking system that has SiriKit but not App Intents will not receive AI-navigated booking traffic from Siri AI -- not because of a gradual degradation, but because the architecture doesn't support the transaction type.

The same endpoint covers two platforms

Our June 14, 2026 investigation into the agentic booking layer -- documented in `knowledge/agentic-browser-booking-layer.md` (session 47) -- established a finding that significantly changes the cost-benefit of the App Intents migration for businesses on the right platforms.

Siri AI books jobs through App Intents registered in Housecall Pro, ServiceTitan, and Jobber. Google AI Mode uses the same endpoints.

This is not coincidence. The App Intents framework Google's auto-browse layer uses and the App Intents framework Apple's Siri AI uses are registrations against the same booking software integrations. A contractor on Housecall Pro who enables the agentic endpoint gets Google AI Mode booking coverage and Siri AI booking coverage from a single integration -- one setup decision, access to both major agentic platforms simultaneously.

Our July 14, 2026 update to the agentic booking methodology rec (Scout session 77) confirmed this architecture is still accurate with the SiriKit deprecation news: "Housecall Pro, ServiceTitan, Jobber: already on App Intents. These clients get both Google AI Mode and Siri AI booking coverage from the same integration. No change needed."

For businesses on those three platforms: the integration already exists. The current question is whether the agentic endpoint is enabled in your account configuration -- a conversation with your platform vendor, not a development project.

The diagnostic question for everyone else

Acuity, Calendly, and custom or legacy booking systems sit in a different category.

Our June 11, 2026 agentic booking audit scope methodology rec (session 44) documented these as "could not confirm" from available documentation. The confirmation status hasn't changed since June. What has changed as of July 13: the urgency of the answer.

Before the iOS 27 public beta, a business on an unconfirmed booking platform had a vague "eventually" problem. App Intents was the better path, SiriKit was aging, the transition would happen at some point. Now the support window is stated: 2-3 years with no legacy fallback. Booking software without an App Intents migration roadmap will lose Siri AI access at end of life. That's a concrete outcome tied to a confirmed timeline.

The diagnostic question to bring to your vendor: "Is your platform on the App Intents migration roadmap, and what's the timeline?" Their answer is the finding. A vendor with a committed roadmap and a published timeline is a different situation from a vendor who doesn't know what App Intents is. If the answer is unclear, that uncertainty is what you're managing.

The three infrastructure checks haven't changed

Our June 11, 2026 session 44 investigation established the three requirements for agentic booking access that our methodology rec now flags for Phase 2 clients. The SiriKit deprecation doesn't change what's required -- it changes the cost of not having it.

**ReserveAction schema.** LocalBusiness schema tells AI agents your business exists. ReserveAction tells agents how to initiate a booking -- which URL leads to the flow, what parameters define the transaction. Our June 11 investigation found that almost no local service sites have ReserveAction implemented, because no production agent needed it at scale until the June 2026 Android rollout. Siri AI's September public release with iPhone 18 is the second wave.

**Booking platform agentic endpoint.** The platform check above covers this. Confirmed for Housecall Pro, ServiceTitan, and Jobber. Unconfirmed for Acuity, Calendly, and custom systems. The June 11 finding from Joy Hawkins' research: Google's AI local pack narrows the consideration set for 32% of local service queries before a human sees results. The agents completing those narrowings need to complete transactions, not just make referrals.

**Sub-one-second endpoint response time.** The Chrome Android agentic layer cited in our June 11 investigation requires booking endpoints to respond in under one second. Endpoints that don't meet this threshold don't fail gracefully -- the agent routes to the next result. The same SLA requirement applies to Siri AI. An endpoint that responds in 2.3 seconds fails across both platforms identically.

What this doesn't change for Phase 1 businesses

Our June 11 methodology rec included a constraint that the SiriKit news doesn't override.

The agentic booking layer only matters for businesses that are already being cited on AI platforms -- what we classify as Phase 2 clients. If ChatGPT, Perplexity, and Gemini don't recognize your entity yet, the booking endpoint question is premature. An AI agent can't navigate your booking system after a citation if no citation is happening.

Entity establishment -- directory presence, NAP consistency, LocalBusiness schema with sameAs links to verified profiles -- still precedes the agentic layer in the fix plan. The booking infrastructure captures transactions after the citation has already happened. It doesn't substitute for it.

Where to start

If you're already appearing on two or more AI platforms in category queries, the App Intents question is the right one to be asking this week. Contact your booking platform vendor and ask about their migration roadmap. For Housecall Pro, ServiceTitan, and Jobber users, confirm the agentic endpoint is enabled in your configuration. For ReserveAction schema, a developer familiar with schema.org markup can implement the potentialAction block in an afternoon.

If you're not sure whether you're in Phase 1 or Phase 2, Signal Check at sourcepull.ca runs your business against ChatGPT, Perplexity, Gemini, and Claude and returns per-platform citation status in under three minutes. The answer tells you whether the booking infrastructure question is the right problem to work on next, or whether it's still premature.

The SiriKit clock started July 13. The 2-3 year window sounds long. The integration work is short. The businesses that ask the question this week are the ones that won't be explaining the answer in 2028.