MacStories puts iOS 27's hundreds of small app fixes ahead of the Siri chatbot

The review holds two verdicts on Siri AI at once, and a roadmap has to pick one of them. MacStories makes the complaint about intelligence. It also says what Siri AI accomplishes with Apple's own apps and compatible third-party ones is "an objectively nicer and more powerful iOS-native experience than anything else on the market right now" ^[11].

Those are two different bets. One is a bet on answer quality, which the review puts below the four American companies it names as leading the AI race ^[8]. The other is a bet on reachability: the proprietary integration with iOS frameworks, personal context and third-party apps that the review calls Siri AI's advantage ^[9]. A feature that only needs to be invoked and handed context works regardless of how well the model reasons. A feature that needs the model to plan several steps depends on it. The review reports the agentic features are missing, and says that will not impress power users who have grown accustomed to Claude Cowork and ChatGPT Work ^[10].

Teams tend to plan the assistant as the main way in. The reviewer, three months in, expects the small app enhancements to mean more to users who care about that sort of thing than "a barebones Siri chatbot app" ^[18].

Availability is the rollout owner's problem. The review's overview attributes the difference between an exciting update and a boring one to geography and defers the detail to a dedicated Siri AI chapter ^[20]. Whether Siri AI's processing happens on the device or on Apple's servers goes unstated ^[19]. For a support team, that means an OS version check settles nothing about whether the assistant a feature depends on exists for the person holding the phone.

The review's own counting puts the weight in the same place. It describes the Siri AI and Apple Intelligence additions as dozens of features, against hundreds of paper-cut fixes ^[15]. Read at the lowest value each word can carry, 24 against 200, the fixes outnumber the AI features by about eight to one ^[17].

The split is testable feature by feature: whether the thing still delivers when the model answers poorly, and whether it does anything at all for a user who cannot invoke Siri AI. Anything failing both tests depends on Apple's model quality and Apple's availability map at once. MacStories says that in eleven years of writing these annual reviews, this was the hardest one to find an angle for. The verdict it landed on is that iOS 27 is not a Snow Leopard release, less flashy than last year's Liquid Glass update and less bombastic for iPad users than iPadOS 26 ^[1][16]. "Everyone wants a so-called Snow Leopard year until it actually happens," the reviewer wrote ^[3].