California: “The Information” claims that Google might reveal its version of Rabbit’s huge action model design as early as December. The project dubbed “Project Jarvis,” is designed to carry out activities for users, such as “gathering research, purchasing a product, or booking a flight.”
Jarvis, which is said to run entirely within a web browser and be optimized especially for Chrome, is purportedly powered by a future iteration of Google’s Gemini.
The tool analyzes and interacts with on-screen features, such as clicking buttons or adding text after collecting screenshots to “automate every day, web-based tasks.” The gadget currently takes “a few seconds” to complete each task.
Also read: Apple Introduces ‘Apple Intelligence’ Across All Devices
Google’s action supports other top AI firms’ initiatives to improve digital task automation. For example, Microsoft’s Copilot Vision will let users talk about the web pages they are now seeing and Apple Intelligence is expected to manage screen-based actions across several apps in the future year.
OpenAI is allegedly developing a model similar to Claude, which Anthropic released in beta form despite being criticized as “cumbersome and error-prone.”
But “The Information” warns that Google’s December preview of Jarvis could still be canceled. The business could decide to release the product to a small number of testers to find and fix any problems.