Google is developing an AI system to automate browser actions

Google is reportedly working on a new artificial intelligence system capable of performing browser actions on behalf of the user. The project, codenamed Project Jarvis, could be unveiled as early as December, The Information reported. Jarvis is designed to perform a variety of tasks, such as “gathering information, buying products or booking tickets,” according to three sources familiar with the project.

Project Jarvis.

Project Jarvis and using Google Gemini

Project Jarvis harnesses the power of the upcoming version of Google Gemini and is customized specifically to work with the Chrome browser. The system aims to automate routine tasks on the web, such as interacting with interfaces, processing screenshots, clicking buttons, and typing text. In its current state, the system performs actions with a delay of a few seconds between steps.

Competition among AI companies

Leading AI companies are also developing similar systems. Microsoft is working on Copilot Vision, which will be able to communicate with the user about the content of web pages. Apple Intelligence plans to introduce features that allow the system to perform actions on the user’s screen in different apps. Also, Anthropic is already testing a “buggy and cumbersome” beta version of Claude capable of running on a user’s computer, and OpenAI is developing a similar tool.

Apple Intelligence Apple Intelligence is developing a similar tool.

Testing and possible timeline changes

The planned unveiling of Jarvis in December may be postponed, The Information reports. Google plans to run the project for a limited number of testers in the near future to identify bugs and refine the system before a mass launch.

So far, Google is planning to run the project for a limited number of testers to identify bugs and refine the system before a mass launch.