Apple Rebuilds Siri with Google Gemini at WWDC 2026

30-Second Summary

Apple used its WWDC 2026 keynote on 8 June to announce the most significant overhaul of Siri since the assistant launched in 2011. The rebuilt Siri runs on a custom 1.2-trillion-parameter Google Gemini model hosted inside Apple's Private Cloud Compute infrastructure, giving it substantially stronger reasoning, multi-step task execution, and personal context capabilities. A new Extensions framework lets users route queries to third-party AI models including ChatGPT, Google Gemini, and Anthropic's Claude, effectively turning Siri into a multi-model interface on every Apple device. The changes ship with iOS 27, macOS 27, and iPadOS 27 in autumn 2026, affecting more than 1.4 billion active Apple devices worldwide.

At a Glance

• Topic: Enterprise AI
• Company: Apple
• Date: 8 June 2026
• Announcement: Apple unveiled a rebuilt Siri powered by Google Gemini at WWDC 2026
• What Changed: Siri moves from a voice shortcut tool to a capable AI assistant that executes multi-step tasks, accesses personal context, and integrates multiple AI providers
• Why It Matters: Every iPhone-carrying employee will have a significantly more capable AI work tool from autumn 2026
• Who Should Care: Business operators whose teams rely on iPhones, iPads, or Macs, and anyone responsible for mobile data and AI policy

Key Facts

• Company: Apple
• Launch Date: Announced 8 June 2026 at WWDC; shipping autumn 2026 with iOS 27
• What Changed: Siri rebuilt on a custom Google Gemini 1.2-trillion-parameter model with multi-step task execution, personal data access, and a third-party Extensions system
• Who It Affects: Any organisation with staff on Apple devices, including iPhone, iPad, and Mac users
• Primary Source: Apple WWDC 2026 Keynote, 8 June 2026

What Happened

Apple announced a complete rebuild of Siri at its annual Worldwide Developers Conference keynote on 8 June 2026, replacing the previous rule-based assistant with a conversational AI powered by a custom version of Google Gemini. The partnership between Apple and Google, first announced in January 2026, sees Apple licensing a 1.2-trillion-parameter Gemini model at approximately $1 billion per year. The model runs inside Apple's Private Cloud Compute infrastructure, which uses Apple Silicon servers with stateless, ephemeral processing. No user data is retained after a query is completed, and Apple's contract explicitly prevents Google from using Apple user queries to train future Gemini models.

The rebuilt Siri ships as a standalone app with a system-wide "Search or Ask" gesture, replacing the previous bottom-of-screen interface. Users can type or speak, attach images and documents, and issue multi-step instructions across apps. Siri can now draft and send emails, update calendar entries, retrieve information from photos and files, and complete cross-app workflows without requiring users to switch between applications manually. The assistant also gains personal context awareness, meaning it can reference a user's recent conversations, upcoming events, and stored documents to provide more relevant responses.

The Extensions system is perhaps the most structurally significant announcement for business operators. iOS 27, iPadOS 27, and macOS 27 will allow users to designate a preferred AI model for Siri queries, with ChatGPT, Gemini, and Anthropic's Claude all supported as third-party options at launch. A system-wide panel triggered by a downward swipe lets users route any query to their chosen provider on demand. Apple frames this as user choice, but for organisations managing fleets of Apple devices, it introduces a new variable: which AI model is handling employee queries, and under what data terms.

This is Tim Cook's final WWDC keynote before he hands the CEO role to John Ternus on 1 September 2026, making it a notable moment for the company's direction. The AI announcements represent Apple's clearest statement yet that it views on-device and cloud AI as central to its product strategy rather than peripheral to it.

Why It Matters

• More than 1.4 billion active Apple devices will receive a substantially more capable AI assistant in autumn 2026, meaning the upgrade is not optional for organisations already on Apple hardware.
• The Extensions system introduces model choice at the device level, which means operators need data policies that cover not just the AI tools their team actively adopts but also the models their phones route to by default.
• Private Cloud Compute sets a new data privacy benchmark: no retention after processing, no training on user data. Business operators can use this standard to evaluate the data handling of every other AI tool in their stack.
• Multi-step task execution on mobile represents a meaningful productivity shift. Tasks that previously required opening multiple apps, copying information, and manually re-entering it can now be completed through a single Siri instruction.
• The multi-provider framework means Apple is no longer betting on a single AI relationship. Operators who have already standardised on ChatGPT, Claude, or Gemini can configure Siri to connect with their existing AI provider.
• The autumn 2026 rollout gives operators roughly three months to update mobile device policies, train staff, and test workflows before the changes arrive on every company iPhone.

The David and Goliath View

Large enterprises will adapt to iOS 27 through their IT departments, MDM platforms, and compliance teams. They will produce policies, training sessions, and approved configuration guides over the next several months. For smaller operators, the risk is the opposite: the changes will arrive unmanaged, with employees routing work queries through whichever AI model they personally prefer, often without awareness that their email content or client files are being passed to a cloud model.

The upside for lean organisations is significant. A team of twenty people, each carrying an iPhone that can now draft follow-up emails, pull contract details from attachments, and schedule meetings without manual steps, has effectively added hours of productive capacity per week without adding headcount. The key is intentionality: operators who decide now which AI model their team should use, what data categories are available to Siri, and which workflows to run through the assistant will capture the benefit. Those who let iOS 27 arrive unmanaged will get the confusion without the productivity.

The actionable recommendation is straightforward. Before autumn, review your mobile device policy and add a section covering AI assistant access to company data. Choose one AI provider from the Extensions list that aligns with your compliance requirements, document why you made that choice, and communicate it to your team with a short practical guide on which tasks are well-suited to the new Siri and which are not. That preparation takes a few hours and turns a platform change into a competitive advantage.

Where This Fits in the AI Stack

AI Growth Engine: The rebuilt Siri can handle multi-step outbound tasks including drafting and sending emails, scheduling follow-ups, and retrieving client context from device data. For sales and client service workflows, this is an automation layer that operates on the device your team already carries.

Employee Amplification Systems: Every employee on an Apple device gains access to a capable AI assistant capable of completing cross-app tasks without manual switching. Organisations that define clear use cases and default settings will see meaningful time savings per person across routine mobile workflows.

Secure AI Brain: Private Cloud Compute's architecture, which guarantees no data retention after processing and prohibits AI training on user queries, represents a meaningful data-handling standard. Operators can use this as a reference point when assessing the data practices of other AI tools in their stack, and should document which AI model is handling Siri queries across their device fleet.

Questions Operators Are Asking

Will my team's data be safe when the new Siri accesses company emails and files? Apple's Private Cloud Compute processes queries on Apple Silicon servers with no data retained after the query is completed, and Google cannot use Apple user queries to train its Gemini models. However, if employees select ChatGPT or another third-party model through the Extensions system, that model's own data terms apply. You should review those terms and set a recommended default model before iOS 27 ships.

Do we need to do anything before iOS 27 arrives in autumn? Three steps are worth completing before the rollout: update your mobile device or acceptable use policy to address AI assistant access to company data, choose a default AI model for the Extensions system that matches your compliance requirements, and identify which employee workflows are well-suited to Siri's new capabilities so you can communicate them clearly when iOS 27 arrives.

Can we control which AI model Siri uses on company-owned iPhones? Mobile Device Management platforms will likely offer controls over the Extensions system for managed devices, consistent with how MDMs currently manage app permissions and data access. If your organisation uses an MDM solution, check with your provider about iOS 27 compatibility and configuration options when developer betas become available.

What kinds of tasks will the new Siri actually handle well for a small business? Multi-step tasks that currently require switching between apps are the highest-value starting point: drafting and sending follow-up emails after a meeting, pulling a client's details from a recent message thread and adding them to a calendar event, or generating a summary of a document from an attachment. Siri's personal context access also makes it useful for retrieval tasks, such as finding a specific email or contract without manually searching.

Does this mean Apple Intelligence is now competitive with ChatGPT and Claude? The rebuilt Siri on Gemini's 1.2-trillion-parameter model closes a significant capability gap compared to the previous version. For everyday business tasks on a mobile device, it will perform substantially better than before. However, for complex, specialised, or long-form work, dedicated tools like ChatGPT or Claude accessed through a browser or API remain the stronger choice. The Extensions system means these are not mutually exclusive; your team can use Siri for mobile task execution and dedicated tools for deeper work.

Citable Summary

What happened: Apple announced a rebuilt Siri at WWDC 2026 on 8 June, powered by a licensed Google Gemini 1.2-trillion-parameter model with multi-step task execution, personal data access, and a third-party Extensions system supporting ChatGPT, Gemini, and Claude, shipping with iOS 27 in autumn 2026.

Why it matters: More than 1.4 billion Apple devices will receive a substantially more capable AI assistant, introducing new mobile productivity capabilities alongside new questions about which AI models are handling company data on employee iPhones.

David and Goliath view: Large enterprises will manage this transition through IT departments and MDM policies; lean operators who proactively define their AI model preference, data boundaries, and priority use cases before autumn will capture the productivity benefit rather than absorb the confusion.

Offer relevance:

• AI Growth Engine: Siri's new multi-step task and email capabilities extend automation to the mobile device layer for sales and client service workflows.
• Employee Amplification Systems: Every Apple device user gains a capable AI assistant for cross-app task execution, multiplying output without adding headcount.
• Secure AI Brain: Private Cloud Compute's no-retention architecture provides a benchmark for evaluating the data practices of all AI tools in an organisation's stack.