Apple Just Replaced Siri With Google Gemini. Here’s Everything That Changes — and When It Arrives on Your iPhone.

New Siri Understands Your Screen, Executes Multi-App Workflows, Maintains Conversation Like ChatGPT, and Learns Your Preferences. The Biggest Upgrade Since Siri Launched in 2011 — Delivered in Two Phases Across iOS 26.4 and iOS 27.

Published: April 8, 2026 | By the Kersai Research Team | Reading Time: ~20 minutes
Last Updated: April 8, 2026

Quick Summary: In January 2026, Apple signed a multi-year agreement with Google — reportedly worth approximately $1 billion per year — to power the next generation of Siri with Google’s Gemini AI. This ends years of Apple struggling to build competitive AI in-house and marks the biggest strategic shift in Siri’s 15-year history. The upgrade is rolling out in two phases: Phase 1 (iOS 26.4, spring 2026) delivers on-screen awareness, contextual understanding, email summarisation, and basic cross-app actions. Phase 2 (iOS 27, September 2026) delivers full conversational AI rivalling ChatGPT — 20+ exchange dialogues, complex multi-app automation, personalisation, and in iOS 27, the ability to use any AI model from the App Store as Siri’s brain. The new Siri runs on Gemini 2.5 Pro — a model with 8x the parameters of Apple’s own AI — processed through Apple’s Private Cloud Compute architecture to ensure Google cannot access your personal data. Compatible devices: iPhone 15 Pro and newer, all iPhone 16 models, iPad with M1+, and Mac with M1+. This guide covers everything: the deal, the features, the timeline, the compatible devices, the privacy architecture, and what it all means for your iPhone — and your business.

Table of Contents

1. Why This Is the Biggest Story in Consumer AI This Year
2. The Apple-Google Deal: What Was Actually Agreed
3. Why Apple Chose Google Gemini Over Building Its Own AI
4. The Privacy Question: Does Google See Your Data?
5. Phase 1: New Siri in iOS 26.4 — Everything It Can Do
6. Phase 2: New Siri in iOS 27 — The Full Transformation
7. iOS 27’s Wildcard: Any AI, Any App Store Model
8. The Release Timeline: Exact Dates and What to Expect
9. Compatible Devices: Which iPhones, iPads, and Macs Get the Upgrade
10. Old Siri vs New Siri: A Complete Capability Comparison
11. What Happened to ChatGPT on Siri?
12. How to Get Early Access via Apple Beta Program
13. What This Means for Australian iPhone Users
14. What This Means for Businesses Using Apple Devices
15. FAQ

1. Why This Is the Biggest Story in Consumer AI This Year

Every AI story in 2026 has been about new models, new startups, new funding rounds. The Apple-Siri-Gemini story is different — because it is not about a new product. It is about something that 2 billion people already use every day, getting its most fundamental upgrade in 15 years.

Siri launched in 2011 — the same year Steve Jobs died, the year the iPhone 4S was announced. For 15 years, Siri has been the punchline of every AI conversation: useful for setting timers and playing music, unreliable for almost everything else. While OpenAI built ChatGPT, Google built Gemini, and Amazon rebuilt Alexa, Siri sat largely frozen — a voice-activated calculator with delusions of grandeur.

The reason was structural. Apple’s AI development operated under a culture of extreme secrecy and internal production — the same culture that produces brilliant hardware but struggles with the collaborative, fast-moving, massive-scale data requirements of frontier AI. Apple built capable AI for specific on-device tasks (photo recognition, keyboard prediction, Face ID) but could not match the conversational and reasoning quality that OpenAI and Google achieved through massive language model training.

The Google Gemini deal breaks that structural constraint. Instead of building conversational AI from scratch — a task that Apple’s own researchers estimated would take years and still might not match frontier quality — Apple licensed the best conversational AI available and wrapped it in Apple’s signature privacy architecture and user experience.

The result: 2 billion Apple devices are about to gain the conversational intelligence of Gemini 2.5 Pro — the same model family that competes with GPT-4 and Claude at the frontier. Not as a separate app. Not as a plugin. As Siri itself.

If you use an iPhone, an iPad, or a Mac — this affects you directly.

2. The Apple-Google Deal: What Was Actually Agreed

Apple officially confirmed the multi-year agreement with Google in January 2026. The broad terms have been partially disclosed through Apple statements, Bloomberg reporting, and market analyses:

2.1 The financial terms

Apple pays Google approximately $1 billion per year for the Gemini integration licence. This is on top of the existing $20 billion annual payment Apple makes to Google to remain the default search engine in Safari — making Google the largest single vendor in Apple’s supply chain by payment, and the arrangement the most financially significant AI deal in consumer technology.

For context on what Apple is paying: Anthropic raised $4 billion from Google and Amazon to fund its entire operation for multiple years. Apple is paying Google $1 billion annually just for AI access — signalling how commercially valuable Apple considers the Gemini integration to its iPhone business.

2.2 The technical structure

The agreement gives Apple access to Gemini 2.5 Pro — Google’s current frontier multimodal model — as the underlying AI engine for Siri’s new conversational and reasoning capabilities.

Key technical points of the arrangement:

• Apple controls the interface: Siri’s visual design, voice, and user experience remain entirely Apple’s. Users interact with “Siri” — they are not directed to a Google product
• Private Cloud Compute: Apple’s secure server infrastructure processes Gemini requests using de-identified data — Google’s servers perform the AI computation, but without receiving identifiable user information
• On-device first: Wherever possible, Siri processes requests on the device using Apple’s local AI. Only queries requiring Gemini’s frontier capabilities are sent to the cloud
• No training rights: Google cannot use Apple users’ Siri queries to train future Gemini models — Apple data stays Apple data

2.3 Multi-year with expansion rights

The deal is multi-year with provisions for Apple to expand Siri’s Gemini integration over time — including the Phase 2 capabilities arriving in iOS 27 and the any-model integration that gives iOS 27 Siri the ability to use any AI from the App Store.

2.4 What Google gets

Beyond the $1 billion payment, Google gains something arguably more valuable: the default AI engine on 2 billion Apple devices. In the competition to become the dominant AI platform, distribution matters enormously. Having Gemini embedded in every iPhone, iPad, and Mac used by premium smartphone customers globally is a distribution advantage that no amount of direct marketing can replicate.

This is why, despite Apple keeping the “Siri” branding and all user interface control, the deal is also a significant win for Google in the AI platform race.

3. Why Apple Chose Google Gemini Over Building Its Own AI

The Apple-Google deal is a strategic admission that raises an obvious question: why couldn’t Apple build this itself?

3.1 The parameter gap

Gemini 2.5 Pro has approximately 1.2 trillion parameters. Apple’s own AI models — which power features like on-device text understanding, photo recognition, and the original ChatGPT integration in iOS 18 — are estimated at around 150 billion parameters. That is an 8x gap in raw model scale.

Scale is not everything in AI — architecture, training data quality, and fine-tuning matter significantly. But for the conversational reasoning quality that makes the new Siri genuinely useful for complex tasks, scale matters a great deal. Apple’s internal models, constrained by on-device deployment requirements, simply could not match frontier conversational quality.

3.2 The time pressure

Apple’s competitive position in the smartphone market is directly threatened by AI quality. Samsung Galaxy phones with Gemini built in, Google Pixel phones with Gemini Ultra, and Android devices generally have been offering significantly better AI assistant experiences than Siri for the past two years. Apple’s premium pricing depends on premium experience — and Siri’s weakness had become one of the most prominent criticisms in iPhone reviews.

Building a competitive frontier AI from scratch would have taken Apple’s AI team three to five more years — years during which Samsung and Android competitors with Google Gemini integration would have had a material AI quality advantage.

The Google deal collapses that timeline from years to months.

3.3 Apple keeps what matters

Apple’s core strategic assets are not the AI model — they are:

• The user experience: How Siri looks, sounds, and behaves remains entirely Apple
• The privacy architecture: Apple’s Private Cloud Compute differentiates its AI handling from every competitor
• The hardware integration: Siri’s on-device capabilities, using Apple Silicon’s Neural Engine, remain entirely Apple
• The ecosystem control: The App Store AI integration in iOS 27 means Apple controls which AI models can power Siri, not Google

Apple has effectively made the same strategic decision it made when it chose to use Intel chips for Macs from 2006–2021: use the best available component from a supplier while retaining control of what matters most, then build internal capability over time to eventually replace the supplier dependency.

4. The Privacy Question: Does Google See Your Data?

This is the question every Apple user is asking — and the answer is more nuanced than a simple yes or no.

4.1 Apple’s explicit commitment

Apple has stated unambiguously: Google cannot access, view, or use your personal data when you use the Gemini-powered Siri. The mechanism is Apple’s Private Cloud Compute (PCC) architecture — a secure server infrastructure Apple developed specifically to enable cloud AI processing without exposing user data to third-party AI providers.

How Private Cloud Compute works:

1. Your Siri request leaves your device encrypted
2. It arrives at Apple’s PCC servers — Apple-operated infrastructure, not Google’s
3. Apple’s PCC de-identifies the request, removing any personally identifying context
4. The de-identified query is sent to Google’s Gemini servers for AI processing
5. Gemini’s response returns to Apple’s PCC
6. Apple’s PCC processes the response and returns it to your device
7. At no point does Google’s infrastructure receive your name, Apple ID, location, contact information, or other identifying data

4.2 The on-device processing layer

For many Siri requests — especially those involving your personal data — Apple processes the request entirely on your device using its own AI models, without sending anything to Google at all.

Specifically, on-device processing handles:

• Queries about your Contacts, Calendar, Notes, and Reminders
• Basic commands (timers, alarms, phone calls)
• Personal context understanding (knowing who “my boss” is based on your contact data)
• Health and fitness data queries

Google’s Gemini is invoked specifically when the query requires frontier-level conversational reasoning, web knowledge, or complex multi-step planning — capabilities that Apple’s on-device models cannot match.

4.3 The honest caveat: metadata

Apple’s commitment that Google cannot access personal data is credible for content — Google does not see the text of your emails, your contact names, or your location. However, as with any cloud AI system, there is always some metadata involved in the communication: request timing, approximate query length, general topic category. Apple has not published a detailed technical breakdown of exactly what metadata passes through to Google’s infrastructure.

For the vast majority of users, this is not a material privacy concern. For users in sensitive professional roles — lawyers, healthcare workers, government employees — understanding the full technical architecture before using Gemini-powered Siri for professional queries is appropriate. Apple’s full Private Cloud Compute technical documentation is available at apple.com/privacy.

4.4 No training on your data

Apple’s agreement explicitly prevents Google from using Apple users’ Siri interactions to train future Gemini models. Your queries do not improve Google’s AI. This is a meaningful protection that distinguishes the Apple implementation from using Google’s own products directly.

5. Phase 1: New Siri in iOS 26.4 — Everything It Can Do

Phase 1 of the new Siri arrives with the iOS 26.4 public release — currently confirmed for spring 2026, with the developer beta running now. This phase transforms Siri from a command-response tool into a genuinely contextual assistant.

5.1 On-screen awareness — the headline feature

The most significant Phase 1 capability is on-screen awareness: Siri can see and understand what is displayed on your screen and respond in context.

This sounds simple. The implications are profound.

Before on-screen awareness:

• You see an article in Safari. You want to know if the author has written anything else. You copy the author’s name, open a new tab, search, navigate to the results.
• You receive an email with a flight number. You want to add the flight to your calendar. You open your calendar, create a new event, type the flight details manually.
• You’re reading a long document. You want a summary. You copy the text, paste it into ChatGPT, request a summary.

With on-screen awareness:

• “Hey Siri, what else has this author written?” — Siri reads the byline, identifies the author, and surfaces other articles.
• “Add this flight to my calendar” while viewing the email — Siri reads the flight details from the screen and creates the calendar event with correct times, flight number, and airline.
• “Summarise this” while viewing any document, webpage, or email — Siri reads the content and produces a summary without you touching the text.

On-screen awareness works across all native Apple apps and many third-party apps that support Apple’s accessibility APIs. It processes screen content on-device where possible, using Apple Silicon’s Neural Engine to read and understand what is displayed.

5.2 Improved context across a conversation

Phase 1 Siri maintains conversation flow across multiple exchanges — remembering what you discussed earlier in a single session rather than treating each request as isolated.

Before: “Siri, call Mum.” → “Calling Mum.” Then two minutes later: “Actually, send her a message instead.” → Siri has no idea what “her” refers to.

After: “Siri, call Mum.” → “Calling Mum.” → “Actually, send her a message instead.” → Siri understands “her” refers to the previous contact and asks what message you’d like to send.

The context window in Phase 1 maintains approximately 5–10 prior exchanges within a single conversation session — meaningful for practical multistep interactions, though not yet the 20+ exchange depth arriving in Phase 2.

5.3 Email and document summarisation

Any email, note, document, or web article visible on screen can be summarised by Siri on demand. This works both for content displayed in native Apple apps (Mail, Notes, Safari) and for content visible in third-party apps with accessibility support.

The summarisation quality draws on Gemini 2.5 Pro’s language capabilities — producing concise, accurate, and nuanced summaries rather than the extractive bullet-point summaries that earlier Apple Intelligence features produced.

Use cases:

• Long email threads: “Summarise this thread and tell me what action is needed from me”
• News articles: “What’s the main point of this article?”
• PDFs opened in Files or Safari: “Give me the key conclusions from this report”
• Meeting notes in Notes: “What were the three main decisions from this meeting?”

5.4 Basic cross-app operations

Phase 1 includes the first generation of Siri’s cross-app action capability — the ability to execute tasks that span multiple applications without your direct involvement.

Phase 1 cross-app capabilities are deliberately conservative — Apple is testing the architecture before expanding scope in Phase 2. Confirmed Phase 1 cross-app actions include:

• “Find the podcast my friend recommended in our messages and play it” — Siri searches Messages, identifies the podcast name, opens Podcasts, and starts playback
• “Take the address from this email and add it to their contact card” — Siri reads the address from the screen, identifies the contact, and updates their record
• “Create a calendar event for the meeting in this email” — Siri reads meeting details from an email and creates a fully populated calendar event
• “Remind me to reply to this email tomorrow morning” — Siri creates a contextual reminder linked to the specific email

5.5 Visual analysis and suggestions

With on-screen awareness extended to photos, Siri can analyse images displayed on screen and provide relevant suggestions or information.

Example use cases:

• A photo of a plant: “What kind of plant is this and how do I care for it?”
• A photo of a dish at a restaurant: “What are the ingredients in this dish?”
• A screenshot of a cluttered desk: “How should I organise this workspace?”
• A photo of a document: “Translate this into English”

This is Gemini’s multimodal capability — its ability to understand images, not just text — applied through Siri’s interface to whatever is visible on your screen.

6. Phase 2: New Siri in iOS 27 — The Full Transformation

iOS 27 launches in September 2026 at Apple’s annual fall hardware event. Phase 2 is where the new Siri becomes what Apple has been promising for years: a genuinely useful AI assistant that rivals ChatGPT and Google Gemini for complex, real-world tasks.

6.1 Full conversational AI — 20+ exchange dialogues

Phase 2 Siri maintains context across 20 or more exchanges in a single conversation — enabling the kind of back-and-forth dialogue that makes ChatGPT genuinely useful for complex problem-solving.

What this enables:

• Multi-turn research conversations: “Tell me about the history of this building” → “When was it built?” → “Who designed it?” → “Are there other buildings by the same architect in Australia?” — Siri maintains the entire context throughout
• Complex task planning: “I need to plan a trip to Japan for two weeks in October” → series of follow-up questions and answers to build a detailed itinerary → “Book the flights” → Siri executes
• Document assistance: Open a complex legal document, ask Siri to explain clause by clause, ask follow-up questions, request a plain-English summary — all in one conversation

6.2 Complex multi-step task automation

Phase 2 introduces Siri’s most powerful practical capability: executing elaborate multi-step workflows across multiple apps with a single natural language instruction.

The example Apple used in its own communications: “Review urgent work emails from last week, create a prioritised memo, set reminders for the top three items tomorrow, and sync to my calendar.”

This single instruction requires Siri to:

1. Access Mail, filter to last week, filter to urgent
2. Read and understand the content of those emails
3. Assess urgency and priority across multiple messages
4. Open Notes and create a formatted memo
5. Identify the three highest-priority items
6. Open Reminders and create three time-specific reminders
7. Open Calendar and create corresponding events

A task that would take a skilled user 15–20 minutes of focused attention — executed by Siri in seconds, with a single spoken request.

Other confirmed Phase 2 automation examples:

• “Find photos from last week’s conference, crop to remove backgrounds, and email to my team with a summary of the event” — accesses Photos, edits images, composes email with context
• “Order breakfast from my usual place, avoid cilantro, deliver to the office, and pay with Apple Pay” — integrates with food delivery apps, applies dietary preferences, processes payment
• “Check traffic on my commute route and suggest an earlier departure time if there are delays” — checks Maps, evaluates conditions against your usual departure time, creates a notification if needed

6.3 Deep personalisation — Siri learns your preferences

Phase 2 Siri builds a persistent preference profile based on your interactions:

• Dietary preferences: Learned from your food orders, notes, and explicit statements — applied automatically to restaurant and food delivery suggestions
• Commute patterns: Learned from your Calendar and Maps usage — used for proactive traffic and departure time alerts
• Communication style: Learned from your email and messages — used to draft responses in your voice when requested
• Work patterns: Learned from your Calendar, task management, and app usage — used to prioritise suggestions and reminders
• Payment preferences: Learned from your Apple Pay history — applied to purchase automations

This preference data is stored on your device and in iCloud, encrypted, and never shared with Google. The personalisation layer is entirely Apple-managed.

6.4 Web research and synthesis

Phase 2 Siri performs multi-source web research and produces comprehensive, cited answers — functioning as a research assistant comparable to Perplexity or the ChatGPT Plus web browsing mode.

Use case: “What are the best-reviewed Japanese restaurants in Sydney under $80 per person, and which ones have availability this Saturday?” — Siri searches multiple sources, cross-references reviews and pricing, checks reservation systems, and produces a ranked list.

This replaces what was previously a multi-app workflow (Google Search, TripAdvisor, Google Maps, OpenTable) with a single Siri query.

7. iOS 27’s Wildcard: Any AI, Any App Store Model

The most strategically significant iOS 27 Siri feature has received almost no coverage — and it may be the most important announcement Apple made at WWDC 2026 (previewed in March 2026 reports).

In iOS 27, Siri will support any AI model offered through the App Store as its underlying intelligence layer — not just Gemini.

This means:

• A developer could publish a Siri-compatible version of their AI model to the App Store
• iPhone users could select that model as the AI powering Siri in Settings
• Anthropic could publish a Siri-compatible Claude model — users could choose Claude as Siri’s brain
• OpenAI could publish a Siri-compatible GPT model — users could choose GPT-4 as Siri’s brain
• Mistral, Llama, local models, specialised models for specific domains — all eligible

This is Apple opening Siri to an AI marketplace — with Apple’s App Store as the distribution and commerce layer. It is the most significant structural development in mobile AI since the App Store itself launched in 2008.

Why this is strategically brilliant for Apple

• Eliminates single-supplier dependency: Apple is not locked into Google Gemini forever. If a better model emerges, users and developers can switch — keeping pressure on Google to maintain quality and pricing
• Creates a new App Store revenue category: AI model distribution is a potentially enormous new category for the App Store, with Apple taking its standard commission on any commercial AI model distribution
• Positions Apple as the platform layer: Rather than competing with OpenAI, Anthropic, and Google at the model layer, Apple becomes the distribution platform for all of them — similar to how the App Store monetised the mobile software ecosystem without Apple needing to build every app
• Responds to regulatory pressure: The EU Digital Markets Act requires Apple to allow interoperability and alternative defaults. The any-model approach elegantly satisfies this requirement while keeping Apple as the controlling platform layer

What this means for developers and businesses

If your business builds AI-powered products for Apple devices — or if you are evaluating AI platforms for business workflows — the iOS 27 any-model architecture changes the calculus entirely.

By September 2026, an enterprise could potentially deploy iPhones configured with a custom AI model — tuned for their specific industry, language, and workflows — as the AI powering Siri for their entire device fleet. A legal firm could use a Siri backed by a legal-AI model. A healthcare provider could use a Siri backed by a medical-AI model. A financial services company could use a Siri backed by a compliance-aware financial model.

This is not confirmed in full technical detail — Apple has not released the full API and developer documentation. But the strategic direction is clear: Apple is building Siri as a universal AI interface, not an AI system.

8. The Release Timeline: Exact Dates and What to Expect

The delay story — what Mark Gurman reported

Bloomberg’s Mark Gurman — Apple’s most reliable internal source — reported in February 2026 that some Phase 1 features originally planned for iOS 26.4 were encountering reliability issues in internal testing. Specifically:

• “Expanding Siri’s ability to access personal information” — cross-referencing data across apps — had delays
• “Siri voice control of in-app actions” — the cross-app automation features — had reliability issues

Apple’s response was to spread features across iOS 26.4, iOS 26.5, and iOS 27 rather than delay a single release. This is a departure from Apple’s traditionally monolithic release approach and reflects the continuous delivery model it has been moving toward since iOS 16.

The practical implication: not every Phase 1 feature described in this article may be available on day one of iOS 26.4. Some features may arrive in 26.5 (May 2026). The full Phase 2 experience is iOS 27 (September 2026) — that timeline is firm.

9. Compatible Devices: Which iPhones, iPads, and Macs Get the Upgrade

Full feature support (Phase 1 and Phase 2)

Limited or gradual support

Why some older iPhones are excluded

The on-screen awareness and complex multi-step automation features require the Neural Engine and processing power of Apple’s latest chips to run efficiently. Older devices can access Gemini via cloud processing for basic conversational queries, but the on-device components that make on-screen awareness work in real time require A17 Pro or newer (iPhone 15 Pro) or M1+ chips.

If you have an iPhone 14 or earlier and this feature set is important to you, iOS 27 this September may accelerate an upgrade decision.

10. Old Siri vs New Siri: A Complete Capability Comparison

11. What Happened to ChatGPT on Siri?

Apple announced ChatGPT integration in Siri at WWDC 2024 as the flagship AI feature of iOS 18. The integration allowed Siri to hand off specific queries to ChatGPT when the question exceeded Siri’s own capabilities.

With the Gemini integration, ChatGPT’s role changes significantly:

• Gemini becomes Siri’s primary AI engine — handling the conversational, reasoning, and task automation functions that ChatGPT previously supported
• ChatGPT remains available as an optional plugin — for users who specifically prefer ChatGPT for certain query types, the option to route queries to ChatGPT remains
• ChatGPT’s role becomes supplementary — rather than a primary capability uplift, it is a user preference option for those with existing ChatGPT subscriptions or preferences

In practical terms: if you never specifically asked to “use ChatGPT” when talking to Siri, the transition to Gemini will be invisible. Siri will simply be significantly better. If you actively used the ChatGPT integration, you will find that Gemini-powered Siri handles most of those queries natively — and ChatGPT remains accessible if you prefer it for specific use cases.

The iOS 27 any-model architecture also means ChatGPT could return as a first-class Siri integration via an OpenAI-published App Store model — giving users explicit choice rather than a single vendor integration.

12. How to Get Early Access via Apple Beta Program

If you want to try Phase 1 Siri features before the iOS 26.4 public release, Apple’s Beta Software Program is free and available to any Apple ID holder.

Step-by-step:

1. Backup your iPhone — go to Settings → [Your name] → iCloud → iCloud Backup → Back Up Now. Beta software can occasionally cause issues; always backup first.
2. Enrol your device — on your iPhone, open Safari and go to beta.apple.com. Sign in with your Apple ID.
3. Enrol in iOS Beta — on the beta portal, find “iOS & iPadOS” and select “Enrol your device.” This downloads a configuration profile.
4. Install the profile — after the profile downloads, go to Settings → General → VPN & Device Management → find the Beta Profile → Install.
5. Update to iOS 26.4 beta — go to Settings → General → Software Update → the iOS 26.4 beta should appear. Tap Download and Install.
6. First boot setup — after the update completes, Siri’s new Phase 1 features are immediately accessible. No additional setup required.

Important notes:

• Beta software may have bugs, reduced battery life, or app compatibility issues
• Install on a secondary device if you rely on your iPhone for business-critical functions
• You can unenrol from the beta at any time in Settings → General → Software Update → Beta Updates → Off

13. What This Means for Australian iPhone Users

Australia has consistently been one of Apple’s earliest international markets for new iOS feature rollouts — and the Gemini-powered Siri is confirmed for Australian English from the initial iOS 26.4 release.

The practical Australian use cases that change immediately

Travel and navigation (Phase 1): “Show me the fastest way to get to [address] from here, avoiding tolls, and add it to my calendar appointment” — combines Maps, Calendar, and Siri’s cross-app capability in a single query.

Business email management: Siri’s email summarisation is immediately valuable for Australian business users drowning in long email threads — especially for replies to international partners in different time zones where context-catching is essential.

Local business research (Phase 2): “Find a good café near my next Calendar appointment that has outdoor seating, check if it’s open, and add it as a stop on my route” — the kind of multi-source local query that currently requires three or four different apps.

The AEST time zone advantage: Because iOS 26.4 public release aligns with the northern hemisphere spring, Australian users will receive the update as Autumn/winter 2026 — a natural time when Australians are often reorganising routines, making the productivity features of Phase 1 Siri immediately relevant.

The Australian privacy consideration

Australia’s Privacy Act 1988 and the Australian Privacy Principles (APPs) govern how personal data can be collected and processed by technology companies operating in Australia. Apple’s Private Cloud Compute architecture — specifically the provision that Google cannot access personally identifiable Australian user data — appears to satisfy the core APPs requirements for cross-border data transfer.

However, organisations deploying Apple devices at scale (corporate iPhone fleets, healthcare iPad deployments) should conduct their own Privacy Act compliance assessment before enabling Gemini-powered Siri features for sensitive use cases. Apple has committed to publishing full Private Cloud Compute technical documentation — reviewing this documentation before enterprise deployment is prudent.

The competitive implications for Australian tech businesses

For Australian businesses building mobile apps or services that integrate with Siri (SiriKit, App Intents), the Phase 1 and Phase 2 capabilities significantly expand what Siri integrations can do. An App Intents integration that was previously limited to basic commands can now participate in complex multi-step workflows orchestrated by Siri.

Australian app developers building in categories that benefit from Siri integration — productivity, health, finance, logistics — should be reviewing their App Intents implementation against Phase 1 and Phase 2 capability documentation now, ahead of public release.

14. What This Means for Businesses Using Apple Devices

The enterprise workflow transformation

For businesses that have standardised on Apple devices — and in Australia, iPhone adoption in professional services, finance, healthcare, and SMEs is extremely high — Phase 2 Siri represents a meaningful productivity capability that requires no new software procurement, no IT deployment complexity, and no training investment beyond familiarisation.

The most immediately valuable enterprise use cases:

Sales and client management:

• “Pull up everything we’ve discussed with [client name] — emails, meetings, notes — and give me a summary of where we are with them” — Siri synthesises across Mail, Calendar, and Notes to provide a relationship summary before a call or meeting
• “After this call, send a follow-up email summarising what we agreed and create tasks in Reminders for my three action items” — executes post-meeting admin in one voice instruction

Administrative efficiency:

• “Summarise all unread emails from this morning, flag anything urgent, and draft a reply to the most time-sensitive one” — processes the morning inbox without manual triage
• “Schedule a 30-minute meeting with [name], find a time that works for both of us next week, and send the invite” — cross-references calendars and handles scheduling

Field workers and mobile teams:

• On-screen awareness means field workers can hold up their phone to a document, form, or piece of equipment and ask Siri contextual questions — transforming the iPhone into a context-aware field tool

The BYOD and MDM consideration

For businesses running BYOD (bring your own device) programs or Mobile Device Management (MDM) deployments:

• Gemini-powered Siri is on by default for compatible devices updating to iOS 26.4 — employees’ personal iPhones will automatically gain these capabilities
• Enterprise MDM controls allow IT administrators to restrict or limit Siri capabilities on managed devices — businesses with strict data governance can prevent Siri from accessing corporate email, calendar, or documents on managed devices
• Private Cloud Compute addresses the primary enterprise concern about cloud AI processing of corporate data — but your IT and compliance teams should verify this against your specific data governance requirements

The competitive productivity consideration

Companies whose competitors adopt Phase 1 and Phase 2 Siri workflows will face a productivity gap with companies that do not. For SMEs — where individual employee productivity directly drives business outcomes — this gap can be material.

The businesses that benefit most will be those that:

1. Proactively identify the Siri workflows most valuable to their specific operations
2. Train employees on the new capabilities (the investment is minimal — a one-hour session is sufficient)
3. Review their Apple device configuration to ensure Siri has the app permissions needed for cross-app workflows

15. FAQ

When exactly is the new Siri coming out?

Phase 1 Siri launches with iOS 26.4 public release — confirmed for late March to April 2026 (the developer and beta versions are already live). Phase 2 full conversational Siri launches with iOS 27 in September 2026. Some Phase 1 features may be spread across iOS 26.5 (May 2026) due to development delays reported by Bloomberg.

Does Google see my personal data when I use the new Siri?

No. Apple’s Private Cloud Compute architecture processes your Siri queries through Apple-operated infrastructure that de-identifies requests before they reach Google’s servers. Google cannot see your name, contacts, emails, location, or other personal data. The de-identified query reaches Google for AI processing, and the response returns through Apple’s servers to your device. Google also cannot use your Siri interactions to train future Gemini models.

How much is Apple paying Google for Gemini?

Reports indicate Apple pays approximately $1 billion per year for the Gemini integration licence — on top of the existing $20 billion annual payment Apple makes to Google for Safari search placement. Apple has not publicly confirmed the exact financial terms.

Which iPhones support the new Siri?

Full Phase 1 and Phase 2 support: iPhone 15 Pro, iPhone 15 Pro Max, and all iPhone 16 models. Limited Phase 1 support: iPhone 15 (non-Pro) and iPhone 14 series. Minimal support: iPhone 13 and earlier. The Neural Engine in the A17 Pro chip (iPhone 15 Pro) and newer is required for full on-screen awareness and local processing features.

What happens to ChatGPT on Siri?

ChatGPT remains available as an optional plugin within Siri but transitions from a primary capability to a supplementary option. Gemini 2.5 Pro becomes Siri’s core AI engine, handling the reasoning and conversational tasks that ChatGPT previously supported. Users who prefer ChatGPT for specific queries can still invoke it. iOS 27’s any-model architecture may allow OpenAI to publish a first-class ChatGPT Siri integration via the App Store.

Can I use a different AI than Gemini with Siri in iOS 27?

Yes — iOS 27 introduces support for any AI model distributed through the App Store as Siri’s underlying AI engine. In theory, users could choose Claude (Anthropic), GPT-4 (OpenAI), or any other App Store-distributed AI model as the intelligence powering their Siri. Full developer documentation is expected at WWDC 2026 in June.

Will this cost extra?

No additional cost for users. The Gemini-powered Siri is included in iOS 26.4 and iOS 27 updates at no charge. Apple pays Google approximately $1 billion per year for the integration, but users receive it as part of their device’s operating system. Third-party AI models distributed via the App Store in iOS 27 may be offered at a cost by their developers — Apple has not confirmed the commercial model for that category.

Is this available in Australia?

Yes. Australian English is a supported language from the initial Phase 1 iOS 26.4 release. Australians with compatible iPhone, iPad, and Mac devices receive the same feature set as US and UK users.

The Bottom Line

Siri just became a different product. For fifteen years, it was the assistant that almost worked — impressive in demos, frustrating in daily use. With Gemini 2.5 Pro as its core intelligence, Privacy Cloud Compute handling your data, and a two-phase rollout delivering conversational AI this spring and full workflow automation in September, the new Siri arrives as a genuine competitor to ChatGPT and Google Gemini in every practical use case.

The iOS 27 any-model architecture is the bigger long-term story: Apple is not just upgrading Siri, it is transforming Siri into a universal AI interface — the layer through which 2 billion Apple device users access any AI, from any developer, through Apple’s platform and distribution machinery.

For iPhone users: update to iOS 26.4 when it arrives and spend 30 minutes exploring on-screen awareness and the new cross-app capabilities. The gap between what Siri could do yesterday and what it can do today is the widest it has ever been.

For businesses with Apple device fleets: the enterprise workflow implications of Phase 2 Siri — available September 2026 — are significant enough to warrant proactive planning now. The businesses that identify their highest-value Siri automation opportunities before iOS 27 ships will capture an immediate productivity advantage.

Kersai helps Australian businesses identify, deploy, and maximise the value of AI tools across their organisations — including the rapidly evolving capabilities of Apple Intelligence, Siri, and enterprise AI platforms. To discuss how your business can prepare for Phase 2 Siri and the iOS 27 any-model architecture, visit kersai.com.

This article was researched and written by the Kersai Research Team. Kersai is a global AI consultancy firm dedicated to helping enterprises confidently navigate the rapidly evolving artificial intelligence landscape. To learn more, visit kersai.com.