Comment by razfar

I'm curious about the computer vision aspect of this tool. Specifically, how was the model which draws bounding boxes around interactable elements trained? Definitely a step beyond existing browser automation software!