Microsoft continues to expand the capabilities of its Copilot app, introducing new features for both a wide range of users and testers. One of the biggest updates is the introduction of image generation directly in the app, powered by the GPT-4o model from OpenAI. This solution replaces the previous DALL-E 3 model, allowing users to create higher-quality visual content without the need for third-party services or additional web tools.
At the same time, a hidden “Action” feature has appeared in the app, currently marked as “coming soon” in the Labs section. It is designed to delegate routine tasks to Copilot for a period of five to ten minutes. Early signs indicate that this feature is primarily aimed at the Windows environment, and its launch will initially be limited—likely to selected testers or Copilot Pro subscribers.
Copilot is also changing its visual style. AI characters, which serve as avatars, are getting a refreshed look. In voice mode, the character now occupies the entire screen instead of a smaller chat window. Notably, the fourth character—still unnamed, resembling a cloud or chewing gum—has become more detailed, although its final appearance has yet to be approved.
The updates are aimed at making interaction with Copilot more flexible and convenient, combining productivity, assistance, and personalization within a single app. Users can already try out the new image generation capabilities, while other features will appear after the testing phase is completed.
