Siri AI Hands-On: Apple’s Revamped Assistant Finally Delivers on the Promise

The new Siri AI, unveiled at WWDC 2026 and now available in developer beta, is receiving its first wave of hands-on reviews. The verdict from early testers is consistent: Apple’s voice assistant, rebuilt on a custom Google Gemini model, is finally useful.

WIRED’s Reece Rogers took the developer beta through its paces in San Francisco, testing it for everyday tasks like restaurant recommendations, navigation, and multi-turn conversations. His conclusion: the new Siri is “conversational, omnipresent, and actually helpful.” Apple has replaced the old circular waveform with a translucent orb that appears at the top of the iPhone screen, spinning while processing requests. Rogers reported that asking for nearby pancake recommendations produced a specific, useful result rather than the generic search results the old Siri would have returned.

Siri AI is not an incremental update. Apple rebuilt the assistant on a custom version of Google’s Gemini model, paying an estimated $1 billion per year under a multi-year deal that could reach $5 billion total, according to analyst Gene Munster of Deepwater Asset Management. The partnership gives Siri access to Gemini’s natural language capabilities while keeping Apple’s privacy architecture on top.

The new assistant supports true multi-turn conversation, on-screen awareness (it can see what is displayed on your iPhone and act on it), and a dedicated Siri app with synced history across devices. The Dynamic Island on newer iPhones serves as the primary interaction point, with the orb animation replacing the old full-screen takeover.

Apple also introduced an Extensions framework at WWDC that will let users set third-party AI models as defaults for Siri and system-wide Apple Intelligence features. Claude, ChatGPT, Gemini, and Grok will all be supported. The framework arrives in public beta in July and launches with the full iOS 27 release in the fall.

The Google Gemini partnership

The Google deal is the centerpiece of the new Siri. Apple had considered building its own large language model but concluded that Google’s Gemini offered the best out-of-the-box performance for the assistant use case. The agreement gives Apple access to Gemini’s reasoning capabilities while keeping user data on-device through Apple’s Private Cloud Compute infrastructure.

For Google, the deal locks in Apple as a major customer for Gemini Enterprise, a significant validation for Google Cloud’s AI strategy. For Apple, it solves the foundational model problem without the years of catch-up it would have taken to build a competitive alternative from scratch.

iOS 27 and the broader picture

Siri AI ships as part of iOS 27, which enters public beta in July. The full release arrives in the fall alongside the new iPhone lineup. Developer betas are already circulating, and early impressions from media and analysts have been overwhelmingly positive.

The assistant works across the entire Apple ecosystem: iPhone, iPad, Mac, Apple Watch, and HomePod. The dedicated Siri app on iPhone provides a scrollable history of past conversations, persistent context across sessions, and the ability to manually trigger the assistant without voice.

Third-party app integrations are handled through Apple’s new App Intents framework, which lets developers expose specific actions to Siri AI. Early partners include Uber, OpenTable, and Spotify, allowing the assistant to book rides, make reservations, and play music without shortcuts or workarounds.

The competition

Siri AI arrives at a moment when every major platform has an intelligent assistant. Google’s Gemini Assistant, Amazon’s Alexa (now powered by Anthropic’s Claude), and Samsung’s Galaxy AI (built on its own Gauss model) are all chasing the same vision of a conversational, proactive assistant that spans devices and services.

Apple’s differentiator is the hardware-software integration that only a vertically integrated platform can offer. Siri AI can control every system setting, access every built-in app, and draw on the Spotlight semantic index for personal context. The on-screen awareness feature, which lets the assistant see and act on whatever is displayed on the screen, is exclusive to Apple’s tight hardware-software integration.

The question is whether these advantages are enough. Google’s Gemini Assistant runs on Android, a platform with 3 billion users. Amazon’s Alexa is in hundreds of millions of homes. Apple’s approach of tightly gating Siri AI to the latest iPhones and iPads limits the addressable market, at least initially.

For now, the early hands-on impressions suggest Apple has finally solved the fundamental problem: Siri works. Whether it works well enough to matter in the broader AI assistant race will be clearer when iOS 27 ships to the public this fall.


Sources: I Let Siri AI Show Me Around San Francisco (WIRED, June 20, 2026); Apple unveils new Siri AI (9to5Mac, June 8, 2026); WWDC 2026 Recap: Siri AI, Gemini Deal (Fello AI, June 2026); I tried Siri AI, and so far it actually works (The Verge, June 2026)

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top