My experiences have been the same, unfortunately. It can do simple tasks, but for anything requiring indirect reasoning or completion of partial content from me

My experiences have been the same, unfortunately. It can do simple tasks, but for anything requiring indirect reasoning or completion of partial content from media (think finishing sonnets as a training content test) Claude just falls flat. Honestly, I'm not sure what makes Claude so "meh". Not to mention having to fill out a Google Doc for API usage? Weird.