That's actually not as bad a take as you'd think; I've been thinking a outdoing this where heavily autonomous agents can ask for feedback from users in exchange for some sats