There’s an old rule that a founder should stop coding as soon as the company can afford someone else to do it. That rule was written when writing software was expensive.

It isn’t anymore.

AI didn’t make developers obsolete — it made the distance between an idea and a working version of that idea almost zero. And when that distance goes to zero, the bottleneck stops being execution and becomes judgment. What’s worth building. What to kill. Which customer complaint is a feature request and which one is noise.

Judgment is the CEO’s job. Execution is the developer’s job. When both live in the same person, and AI handles the volume, the company runs on one feedback loop instead of five.

What that looks like in practice:

A customer flags a problem on Tuesday. I don’t write a ticket, wait for grooming, negotiate scope, and see it three sprints later. I understand the problem — because I talked to them — and I build the fix, because I know the codebase. AI writes the boring 80%. I own the 20% that matters and the decision to ship it at all.

What it doesn’t mean:

It doesn’t mean one person replaces a team. It means the team is smaller, more senior, and spends its time on the parts that don’t compress. AI is very good at the work between the idea and the artifact. It is not good at knowing whether the artifact should exist. Confusing those two is how companies ship a lot of things nobody asked for, faster than ever.

The bet:

The next generation of companies won’t be split into “technical founders” and “business founders.” That split was a workaround for a cost structure that no longer exists. The operators who win will be the ones who can hold the whole loop — market, product, code, revenue — and use AI to carry the weight in the middle.