Vy is a native Mac app powered by combining our model's advanced interaction capabilities with frontier reasoning agents. It runs directly on your machine. It works with your actual software, on your screen.
I attempted to automate the creation of a LinkedIn banner using Vy by integrating it with Canva. While the tool successfully navigates to Canva and initiates the banner creation process, it struggles with reliably selecting and replacing text elements. It often attempts to change the name but then prematurely jumps to the next task, leaving the original text blank or incomplete. Additionally, Vy appears to take considerable time to interpret what’s happening on the screen—such as capturing and understanding screen elements—which slows down the overall automation process. To improve performance, it would be beneficial if Vy could incorporate not just UI-based automation but also background-level automation. This could enable faster decision-making and smoother task execution. Overall, I see immense potential in the tool. However, it still needs refinement in areas like automation speed and fluid tab navigation within a browser.