Everybody is a Manager

Almost 17 years ago, Paul Graham wrote an influential blog post comparing maker’s schedule vs manager’s schedule. The idea is that makers need blocks of uninterrupted time to think and code while managers have small blocks of time for meetings with different people.

This is very accurate. When I was a manager at Google or ran my startups, my schedule was packed with half hour long meetings with people with all kinds of different backgrounds and tasks. I always start my day way earlier than everybody else because that’s the only time I can find long uninterrupted blocks for myself to think, to reflect and to make.

However, neither Paul nor anybody else could have predicted that AI has turned everybody into managers. AI is so efficient and prolific at writing code and making designs, human experts have shifted into AI’s manager, ie. providing guidance, reviewing AI’s work and making decisions. Unlike a traditional manager with 5 to 10 direct reports, one can have essentially unlimited AI agents, and each of them works so damn fast and non-stop. This is going to make every human a very exhausted manager.

This has been noticed. Siddant has written about his experience as a software developer, but it’s definitely not unique to software engineering. Designers could spend more time reviewing AI outputs than actual designing. Authors could spend more time reading AI writeups than actual writing. Full disclosure, this post is written purely the old fashion way; no AI was involved or harmed during the writing process. As more and more AI is adopted, the tsunami of AI outputs is going to be overwhelming for us.

One word of advice as I always tell anybody who has recently become a manager: you should set a block of time uninterrupted for yourself to think and to reflect. These AI workers can wait, you need the uninterruption to learn, to reorient, to provide guidance and be more effective.