How I Divide the Labor

This week I gave a keynote on what I have been calling the new division of labor. The argument: AI was supposed to handle certain kinds of work so we could focus on others. Instead we outsourced the thinking first, because it was the most effortful, and the result has been cognitive dependence and the slow erosion of the judgment that made us valuable in the first place. I am writing a fuller piece on that argument. This is the practical companion.

A core question that kept coming up: what does this division actually look like?

Throughout my day, the division looks like this.

Every conversation is transcribed by Granola running in the background. I am not thinking about documentation. I am in the conversation. By the end of the day I have a complete record of every idea, every commitment, every nuance, data I can swiftly act on and format to move things forward.

Writing, research and analysis happen in Claude, with custom skills built around my voice and frameworks. This is not “write me a blog post.” It is a system I think with.

Deck-generation is handled there too (and it’s marvelous). My role is the argument, the carefully thought-through outline.

Scheduling is handled by Howie. Meanwhile, automations that would take months to get to are being built by Claude CoWork, laying the systems for distribution of my work.

Each of these is a division of labor in miniature: transcription, thought partnership, scheduling, autonomous execution on one side. Judgment, voice, perspective, decisions on the other. None of them think for me. In theory. In practice the interfaces are designed to draw you further in, to propose their thinking, to offer the next step before you have taken your own. Retaining your position in that relationship takes constant vigilance. But when the division holds, they open up the capacity to think and give me intelligence to work with that I did not have before.

The temptation is always to let these eloquent thinking machines go further, to let them think through the implications before you have had a chance to form your own view. Each temptation not resisted takes you further from your own thoughts, your most valuable asset.

The test is not how much you produced. The test is time to insight. Are you arriving at clarity faster? Are you spending more of your day on the thinking that actually matters? If yes, the division is working. If you feel like an appendage to the machine, it is working against you.

More on what this means, why it matters, and where it comes from, soon.