The hardest AI decision is where not to use it.

Not every problem is an AI problem. Some work is worth money precisely because a human does it: voice, judgment, relationships. Put a model there and the only thing that mattered is gone.

I build with AI all day, and the discipline that makes that safe is knowing where to stop. I have killed projects of my own rather than let AI become a rescue narrative for ideas that needed smaller, real tests first. I keep AI out of my own voice, my read on a person, and my relationships. This page is built the same way: no chatbot, no generated voice.

Before funding anything with “AI” in the title, name the parts where it would be theatre, and protect them on purpose.