Field Note
We spend a lot of time talking about the people building AI.
I think we're overlooking the people who are going to operate it.
Not necessarily software engineers. Not necessarily data scientists.
People who understand how real work happens well enough to direct AI through it.
They know where judgment is required. They know when a process breaks. They know which exceptions matter. They know when something that looks right on paper is completely wrong in practice.
Those people may become far more important in the AI economy than we realize.
Because not everyone needs to build AI.
We are going to need millions of people who know how to direct it.
Think about someone who has spent 10, 15, or 20 years inside a business.
They don't just perform tasks.
They know the shortcuts. The exceptions. The bad data. The unofficial process. The thing everyone does because the official process doesn't actually work.
Most of that knowledge is never documented.
And as companies rush to automate, there's a risk they remove the person before they capture what that person knows.
I think the better opportunity is to turn experienced people into AI operators.
Teach them how to break work into steps.
How to direct agents.
How to evaluate outputs.
How to create rules, approvals, and escalation points.
How to know what AI should handle and what should stay human.
And I think it could become a real trade.
A good AI operator doesn't need to know how to build the model.
They need to understand the work well enough to supervise the system doing it.
That's part of the thinking behind Skills for AI.
Take someone with deep operational experience and teach them agents, workflows, evaluation, and orchestration.
Then let AI multiply what they already know.
Because automating someone's tasks is not the same thing as replacing their knowledge.
Before companies automate experienced people away, they should ask a much better question:
What does this person know that our systems don't?
The people who understand how work actually gets done may not be the workforce AI replaces first.
They may be the people AI needs most.