Don’t solve today’s problems.

That’s an exaggeration, but there’s a real trap here. Don’t spend weeks solving a problem that the next AI release is likely to solve for you. At the same time, don’t build for a problem you think you’ll have a year from now. Too much is going to change between now and then.

Solve tomorrow’s problems. Work on the things that are just beyond what AI can do for you today, but close enough that you know you’ll actually need them. In a world changing this quickly, that’s about as far ahead as I want to build.