Has the iPhone version of Doubao Mobile Assistant appeared? Developer demonstrates operating an iPhone with Opus 5.5.

Beating AI News Flash: Developer Finn Voorhees demonstrated an AI Agent that can directly operate an iPhone. He simply sent one sentence, asking the Agent to find the cheapest Ryanair flight to London for the next day. The Agent then opened Ryanair on its own, entered the destination, selected the date, checked fares, and finally returned to the chat interface to report the results. No computer was involved throughout the process.

This Agent runs in PiChat, an iPhone chat app developed by Finn. Under the hood, it uses PiDurableKit, which he previously open-sourced, to run the Pi Agent on the phone through Apple's JavaScriptCore, handling model calls, tool execution, and task management. The video uses Claude Opus 5.5, with model inference still in the cloud. To allow the Agent to continue working after switching to other apps, PiChat also uses Apple's background task mechanism and saves execution progress via SQLite, making it easy to resume after interruption.

The most critical cross-app control code has not yet been made public. In the video, when iPhone Control is enabled, the system requires entering the lock screen passcode to enable XCTest UI automation. XCTest is Apple's interface testing framework, capable of simulating taps, swipes, and text input. The open-source project TouchSynthesis uses XCTest's private interfaces to let the iPhone generate touch events on its own, without the need for real-time computer control. However, Finn has not yet confirmed whether this solution was used.