Google plans to present the new AI project under the code name “Project Jarvis”, reports the Information with reference to sources familiar with the development.
A system operating on an updated version of the Gemini model will be able to independently perform various tasks in the Chrome browser: conduct research, make purchases and book airline tickets.
The principle of operation is based on the analysis of screenshots of web pages, followed by pressing buttons and input text. According to sources, in the current version, each action takes a few seconds.
Other technological giants lead similar developments. Microsoft creates Copilot Vision to interact with web pages, Apple is working on the Intelligence system to manage the applications, and Anthropic and Openai develop computer management technologies through AI
According to the publication, Google plans for demonstrating Jarvis may change in December. The company considers the possibility of releasing a system for a small number of testers to search and eliminate errors.