The AI Monetization Fantasy

Everyone wants to sell the AI thing they haven’t built yet.

A department, a division, an enthusiastic colleague, convinced the tool they’re about to create will become a new revenue stream. They’ll build it, sell it to other hospitals, and watch the money flow.

Derek Sivers made the point years ago: an idea is worth almost nothing, execution is worth almost everything. I picture it as a staircase. The idea is step one. Making something that works, step ten. Getting another organization to run it, step one thousand. Most people stand on the first step, already shopping for bags to carry the money.

They skip adoption. It’s one thing to make something useful for your team. It’s another for your team to actually use it, to find it faster than the old way, and to complain if you took it away. That gap, between “we built it” and “we can’t work without it,” is where ideas go to die.

And if you built your solution with the same off-the-shelf AI tools as everyone else, your solution itself will become a commodity. Same tools, same outputs. There is no moat.

Build something. Get people to use it. Prove it matters.