You are probably right to the extent of multimodal for some tasks or for sensor input. But redlining a 200 page set of drawings to capture as built info is still different from assembling that set of drawings in the first place. I doubt it will be all or nothing. E.g., supporting field service will require generating particular views, but I don't see engineering review meetings skimping on monitor pixels, at a cost of N laptops per hour just to assemble a meeting.