Source: Nate’s Substack
The productivity paradox of autonomous agents is real: by lowering the friction to initiate tasks, they've inverted the constraint from execution to decision-making. Users now face decision bottlenecks—approving, editing, or triaging agent outputs—that didn't exist when task initiation itself required friction. Agent tools are not time-savers for knowledge workers, but force multipliers that require new management disciplines, which many users haven't developed or resist adopting.