I considered using something like this, but as I explained in another comment (https://news.ycombinator.com/item?id=21254984), I think visual matching beats other approaches. I think the few seconds it takes to think about which app owns the window, or what text could match the window, will be longer that directly looking and visually recognizing your window. Again, this opinion is not supported by research. You may want to experiment on yourself. It is also not only about speed but the comfort of each work-flow may differ for different people.