If you're the human holding judgment in an AI-native company, you've already switched roles — and your old scheduling habits will quietly wreck your throughput anyway.
Paul Graham's essay draws a clean line between two incompatible rhythms: managers slice their days into one-hour appointments and feel productive doing it, while makers — programmers, writers, anyone in deep construction mode — need half-day minimums just to reach working depth. A single meeting dropped into the middle of a maker's afternoon doesn't cost one hour; it costs the whole block, because the anticipation alone prevents real immersion. Schedule design, Graham argues, is output design, not calendar hygiene.
Your agents are now the makers. They don't attend your 11 a.m. check-in, but you do, and that's the problem. Every time you pull yourself into a coordination meeting, you interrupt the only thing agents genuinely need from you: coherent, considered judgment on ambiguous decisions. Shallow, fragmented attention produces shallow directives, and agents will execute shallow directives with impressive thoroughness. The founder's job in an AI-native studio isn't to stay busy; it's to stay deep enough that when judgment is required, it's actually worth something. Protect the blocks accordingly.
- Guard your deep-work blocks as if agents bill by the quality of your decisions, not your hours
- design your day so coordination tasks cluster at the edges, leaving the interior for the judgment calls only you can make
- recognize that a fragmented founder becomes a bottleneck dressed up as a manager.
