Google is reportedly preparing to unveil its own version of Rabbit’s large action model concept, potentially as early as December. According to The Information, this initiative, codenamed "Project Jarvis," is designed to perform tasks for users such as compiling research, purchasing products, or booking flights. This information comes from three individuals with direct knowledge of the project.
The tool is expected to be powered by an upcoming version of Google’s Gemini and is intended to function exclusively with a web browser, specifically optimized for Chrome. Its aim is to assist users in automating routine, web-based tasks by interpreting screenshots and performing actions like clicking buttons or entering text. The Information notes that the current version of Jarvis takes a few seconds to execute these actions.
However, The Information also highlights that Google’s plan to introduce Jarvis in December could be altered. The company is contemplating releasing the tool to a limited group of testers to identify and resolve any bugs.
/cdn.vox-cdn.com/uploads/chorus_asset/file/25263501/STK_414_AI_A.jpg&description=Google+Developing+AI+System+as+a+%E2%80%98Computer-Using+Agent%E2%80%99){kind=link}