Deep research is still not very good and can be improved (will be better tomorrow than today), but the vector of improvement is clear and guaranteed, making it a tool worth adopting despite current limitations because the trajectory is known and positive.