Saw some discussion of people using LLMs to assist with pairings in team tournaments. Not only is this stupid and wasteful, but it's *the wrong type of model*. You could actually implement a pairings model. Would take maybe a weekend worth of coding. Be methodologically correct in your cheating.