Agreed. Color me skeptical. All of the interactions and decisions described are plausible, but in my experience with AI agents, they would require frequent human intervention.