Operator for iPhone

Inspired by OpenAI’s “Operator,” our system takes the idea further—harnessing the power of multiple autonomous iPhones to execute complex, human-like tasks in parallel.

One Thousand iPhones In Your Pocket

Break down any task into "n" subtasks, each executed by a virtual iPhone agent acting as a full, self-contained operator.

These agents aren’t limited by App Intents or third-party APIs—they can do everything a human can do on an iPhone: navigating apps, adjusting system settings, tapping, swiping, typing, and more. The result is a radical leap in phone-native automation.


Distributed Task Execution

Complex tasks are decomposed and intelligently dispatched across a fleet of autonomous iPhones (real or virtual), each handling a portion of the workload simultaneously. Think: “MapReduce for iOS.”

Full iOS Control

Agents can operate across the entire iOS surface—not just apps with intent support. They mimic human behavior directly, offering complete automation without requiring any developer buy-in.

Image-Based Feedback Loop

Each iPhone streams UI snapshots to a custom-built server which feeds visual data into Claude (Anthropic) for interpretation, enabling high-accuracy state awareness and dynamic decision-making.

Custom Hardware & Demo Design

For our internal launch demo, I engineered a rig of twelve real iPhones operating simultaneously—delivering a tangible, high-impact visualization of the system at work. This real-world demo emphasized our system’s power and fidelity in a way that virtual phones could not.


My contributions

Designed and orchestrated the live demo, coordinating the hardware rig and the task distribution framework

Built the server-side infrastructure that processed and routed image-based feedback from devices

Worked closely with our design and systems teams to ensure that the visual, interactive experience of the demo aligned with the core product vision

Developed the physical housing and power/distribution system for showcasing real devices in a tightly choreographed presentation