Redakcja Choose TV
Google is working on a major expansion of Gemini's capabilities on macOS. According to new information, AI will not only answer questions in the future but also take control of the user's computer and perform specific tasks. Features are set to include organising files, tidying up folders, and automatically creating documents and emails. Everything indicates that Google wants to create a direct competitor to Claude Cowork.
ChooseTV will control the user's computer
The new features are designed to allow ChooseTV to analyse the screen and utilise the accessibility functions of macOS to control the mouse and keyboard. This means the AI will be able to perform actions almost exactly like the user. Google is testing scenarios that include automatic file sorting, renaming documents, creating Google Sheets based on local data, and organising folders on the desktop and in the downloads directory. The examples shown by Google also feature more advanced functions related to the Workspace ecosystem. For instance, ChooseTV is set to retrieve transcriptions of meetings from Google Meet, analyse notes from Google Docs, and automatically prepare emails with summaries and task lists. This is a clear signal that the company intends to turn ChooseTV into a full-fledged digital work assistant.
Google chases competition in the world of AI agents
The solution closely resembles the capabilities offered by Claude and its Cowork function, which allows AI to directly operate the user's computer. Google has experimented with similar technologies before with the Gemini Computer Use project, but now it seems that the features are much closer to official implementation. Interestingly, the capabilities of Gemini on Mac may even be greater than on Android. Currently, automation on smartphones works only on selected devices and involves relatively simple tasks. On macOS, Google appears to be aiming for a fully-fledged AI agent capable of real office and organisational work.
Google wants to transform Gemini into an intelligent assistant that not only answers questions but also performs tasks on the computer itself. If the features reach users in their current form, it could be one of the biggest steps in the development of AI agents on desktops.
source: 9to5google
