How AI agents could remake how iPhone users do everyday tasks is shown in a new open source project. The tool, developed by Rounak Jain, uses OpenAI’s GPT 4.1 to execute complex multi step commands that normally require multiple manual actions. This free prototype from an OpenAI hackathon in 2024 is not on the App Store but you can download it for free on GitHub.
This AI agent goes a step beyond traditional virtual assistants which need user confirmation or only give simple suggestions, by doing things on its own. To illustrate, if I give the command, “Text John that I’m on the way and book an Uber to his place”, the AI will open Messages, send the text, switch to Uber, book a car and not need further commands. This is a huge leap in user convenience of multi app workflows that remove friction.
The AI agent can currently operate within Apple’s automation friendly framework by using Apple’s built in app Shortcuts to interact with apps by using preconfigured workflows. Apple is very strict with app sandboxing rules and privacy policies also restrict technically how deep third party apps or agents can integrate with system level functionality. Therefore, Jain’s project is more of a proof of concept than a finished consumer facing tool.
Notwithstanding, the concerns it raises are fundamental ones in terms of digital assistants over the long haul. The promise is they will expand vastly how users interact with their devices, allowing smart automation, hands free productivity, automation and a very smooth and effective way to transport information from app to app. This could therefore also force companies such as Apple to allow more flexible integrations of apps or even introduce their own smart AI agent in future iOS releases.
Jain’s project suggests near future when artificial intelligence becomes your personal agent that proactively simplifies your life, interpreting your intent rather than instructions; potentially transforming how we use smartphones once and for all.
