Any text-centric action will have a great advantage when done inside a terminal. I would think the equivalent in GUIs would be first-class OCR support on the compositor level. I am not informed about any progress made in that region to be honest, so I cannot tell how close we are (or not) to this.