Here's an interesting lens for evaluating AI's impact: think about what AI actually does at its core.

AI agents are fundamentally built to replicate agency—the ability to make decisions, take action, and adapt independently. That's the whole point.

So here's where it gets interesting: jobs fall into two categories when you look through this framework.

First, there are roles that fundamentally require agency—judgment calls, adaptive problem-solving, creative direction, strategic decision-making. These aren't tasks that follow a fixed script. For roles like these, AI becomes a force multiplier. One person can now coordinate and oversee what previously required a whole team. The work still happens; it just scales differently.

Then there are jobs that don't really need agency—repetitive tasks, data processing, pattern matching, routine execution. These are exactly what AI was built to handle. When a role is purely mechanical, automation doesn't augment it; it replaces it entirely.

The real question isn't whether AI will change work—it will. It's which roles sit in which bucket, and how quickly companies figure that out. The gap between those two categories? That's where all the actual change happens.