Taste Is the Moat

I've built a descent to the bottom of the ocean. Also a launch film for a concept car, made by a carmaker that doesn't exist. They live in my lab, labelled exactly what they are, invented brands running on real code. A reasonable person might ask why a founder spends serious hours on elaborate software for imaginary companies.

Short answer: the AI era quietly repriced every skill I have, and this is the one that appreciated.

Execution collapsed. Selection didn't.

Building is cheap now, genuinely cheap. I direct AI through most of my execution and the distance from idea to working software keeps shrinking. Which means the value went somewhere else: into knowing which thing to build, what it should feel like, and whether it's actually finished or just stopped.

Those are taste questions, and taste has one property that makes it a moat. You can't prompt for it. The model hands everyone the same enormous average. What it can't hand you is a point of view about which part of that average is wrong.

When everyone can build anything, what you choose to build becomes the entire signal.

Concept work is taste training

You don't develop a point of view by reading about one. You develop it by making a thousand specific calls and then living with them. Concept work is where I get the reps. HADAL forced a question about restraint: how slow can a descent be and still hold someone's attention? VEYLO forced coherence, because inventing a car brand means deciding what it would never do, which is exactly the muscle you need for deciding what your real company would never do.

Invented brief, real constraints, real finish quality. The brands are fictional. The judgement compounds into everything I build for real, so the ventures inherit whatever the lab trained.

Label your fiction

One rule I hold firmly: concepts get labelled as concepts. Everything in my lab says so, plainly. Partly that's ethics, since the internet has enough fabricated portfolios posing as client work and I'd rather not add to the pile. But it's also strategy. "I invented this brand from nothing and built it to this standard" is a stronger sentence than any pretend engagement. The label doesn't diminish the work. It's the work's best credential.

Honesty and taste turn out to be the same discipline at different scales. Both come down to refusing the convenient version.

The portfolio inversion

A portfolio used to answer one question. Can this person build? That question is on its way out, because soon everyone can build, or can direct something that builds. The question a body of work answers now is different. What does this person choose when nothing is forcing their hand? What do they refuse? What do they finish, and to what standard?

Which is why the lab sits next to the ventures on this site without apology. The ventures show what I've committed to. The lab shows how I decide. In an era of infinite capability, the deciding is the résumé.