Share with your friends!

Apple used its Worldwide Developers Conference keynote to formally introduce the long-delayed AI overhaul for Siri, promising a more conversational assistant that the company is now calling “Siri AI” and saying the update will arrive in OS releases “this fall.” The new version is part of a broader Apple Intelligence push that also includes an updated on-device Foundation Models stack and deeper integration across Apple’s operating systems.

Apple’s pitch: useful AI, not AI for its own sake

In the keynote, Apple Software Engineering senior vice president Craig Federighi drew a contrast between Apple’s approach and rivals that, as he put it, “appear to be racing forward, seemingly pursuing AI for the sake of AI, with little regard for the people… it’s meant to serve.” Apple, he said, believes “that truly helpful AI must be centered around you and your needs.”

That theme ran through a set of scripted demos meant to show Siri AI as a more natural, multi-step assistant rather than a tool for isolated commands. The demonstrations featured long pauses between questions and answers, underscoring the assistant’s new conversational flow and its ability to switch between different tasks and apps as needed.

Multi-step tasks across apps and data

Apple’s examples showed Siri AI doing more than answering a single question. In one demo, a user asked for World Cup schedule details, then requested recipes inspired by a Brazil vs. Morocco match, then asked for a dessert recently mentioned by a friend named Maria in Messages. Siri found the dessert reference, assembled the information into a watch-party menu, and sent that menu to a group chat along with an invite.

Another demo began with a question about where a photo of an arch had been taken. That led to finding the address of a friend named Jeff who had recently moved. With that context, Siri was able to provide directions in Apple Maps to the arch with a stop at Jeff’s home, without the user manually piecing together the information.

Apple said the system is designed to use “personal context” from data on the device, allowing Siri to search across messages and email even when users can’t remember where something was stored. The company also said that when writing emails, Siri can tailor its tone and style to match a user’s prior messages to the same person.

Web knowledge, app actions, and on-screen awareness

According to Apple, Siri AI also taps “world knowledge” by searching the web through private cloud compute to generate answers. The assistant can use app actions to determine which tools are needed to complete a task, and it can understand what is currently on screen to make its responses more relevant.

Apple also highlighted what it called “visual intelligence.” In one example, Siri could answer questions directly from the camera app. In another, it extracted multiple calendar items from a concert schedule image. On VisionOS, users will be able to ask questions about physical objects in the world around them and get responses from Apple Intelligence.

Apple said a new “Write with Siri” feature will let people generate text “virtually anywhere you type.” A new AI-powered proofreading system will also check writing style system-wide, in a way Apple compared to spell check.

Where Siri AI will show up

Apple said the Siri AI experience will be woven into multiple platforms:

macOS: Siri AI will be integrated into Spotlight, which can decide when a typed query should become a Siri conversation instead of a file or web search. Users can also ctrl-click across the OS to ask about photos, files, text, or multiple files at once.
VisionOS: Apple showed a glowing, animated Siri Orb that sits in the workspace and responds whenever users look at it.
iCloud and a new Siri app: Siri conversations will be stored locally and via iCloud in a dedicated Siri app, letting users start on one Apple device and continue on another.

Two tiers of Apple Intelligence

Apple also used the keynote to clarify that not every device will get the same level of AI capability. The company announced a two-tier structure for its on-device Apple Intelligence features. Its “most capable model” will be limited to devices that meet specific chip and memory requirements:

iPhone Air and iPhone 17 Pro
iPads with M4 and later CPUs and 12GB or more of memory
Macs with M3 and later chips and 12GB or more of memory

Devices that already support Apple Intelligence but do not meet those requirements will still get a less capable model, but it will lack some features. Apple said that includes a new, more expressive Siri voice, which can be adjusted with a slider for both expressiveness and speaking pace. The more powerful model also brings what Apple described as a major improvement in dictation accuracy, with better spelling and punctuation.

Many Apple Intelligence features will still depend on private cloud compute. Apple also said image generation and other off-device features will be limited by daily usage caps, with expanded usage available to subscribers on most iCloud+ plans.

Language and privacy

Apple said the new Siri AI features will launch in English first and expand later to every language supported by iOS. The company again emphasized that the system was designed with “non-negotiable” privacy protections, saying Siri AI conversations will not be accessible to Apple or anyone else.

Apple’s WWDC presentation suggested that the company wants Siri to become a more personal, context-aware assistant that can handle longer requests, combine information from multiple apps, and adapt to what users are doing in the moment. Whether that promise translates as smoothly outside Apple’s demo stage will be one of the big questions when the update arrives later this year.

Source: Original report