According to The Information, Google may showcase its interpretation of Rabbit’s large action model by December. The project, reportedly named “Project Jarvis, ” aims to assist users with tasks such as “conducting research, buying products, or booking flights, ” based on insights from three individuals familiar with the initiative. Jarvis is expected to operate using a future version of Google's Gemini and is specifically designed for use with a web browser, particularly Chrome. The tool is focused on helping users “automate daily web-based tasks” by taking screenshots, interpreting them, and performing actions like clicking buttons or entering text, as reported by The Information. Currently, it reportedly experiences a delay of “a few seconds” between actions. Leading AI companies are concurrently developing models that align with what The Information describes.

Microsoft’s Copilot Vision will enable users to converse about the webpages they are viewing, while Apple Intelligence is anticipated to recognize on-screen content and execute tasks across various apps over the next year. Anthropic has introduced a “cumbersome and error-prone” beta update of Claude that can operate a computer on behalf of users, and OpenAI is also believed to be working on a similar capability. The Information adds that Google’s timeline for unveiling Jarvis in December may change. The company is considering a limited release to a small group of testers to identify and resolve any bugs.