automation-first learning

AI is merging learning and building into the same act.

Where before we would read a lot -> understand a domain -> then build something with that knowledge.

Now we start building in a domain we don't fully understand, and the build itself becomes our knowledge. The automation you create is both the learning artifact and the tool, each iteration chips away at the unnecessary as you develop intuition through contact with real constraints, not theory.

LLMs reduce the gap between "I wonder if this would work" and actually testing it. You no longer need domain expertise just to start, so now thinking naturally wants to become execution, something tangible, not just ideas sitting in your head. It's like your brain grows tendrils that reach into reality and act, dissolving the boundary between thinking and doing.

The world is like legos now, everything is composable raw material you reshape as your understanding deepens.

Most AI discourse is about productivity ("do things faster") but I think the actual shift is more fundamental: the nature of the activity changed; learning, building, and toolmaking are becoming one act.

How do you build for this? Go touch reality with your brain.