Rick Rubin’s philosophy of creating something truly for yourself is an inspiration for this AI era.

“The audience comes last… I’m not making it for them. I’m making it for me. And it turns out that when you make something truly for yourself, you’re doing the best thing you possibly can for the audience.”

That’s a strange standard to hold software to. Software was the opposite of personal art. It got built by teams, for markets, on roadmaps, with the actual user somewhere downstream adapting to whatever shipped. You didn’t make your tools. Most of the time, you filed feature requests and waited.

That changed this past year. People call it vibe coding: you describe what you want in plain language, the model builds it, you look and say yes or no, and you go again. The loop is quick, and you don’t need a team or anyone’s permission. An audience of one, building for an audience of one.

It works because the loop runs on recognition. You don’t have to know in advance what you want. You only have to recognize it when it shows up, then steer. Almost anyone can do that, including people who could never write the code themselves.

And the model never tries to take the wheel, because it wants nothing. It was trained on everything that already exists, so it can hand you the average of every app ever made, instantly. The one thing it can’t do is be dissatisfied. It has no quarrel with what exists, because what exists is all it has seen. The itch that none of the current tools is quite right has to come from you.

A book I read recently and loved is Jordan Mechner’s The Making of Prince of Persia: Journals 1985-1993. Through the late 1980s he built the game almost entirely alone. To make his little running figure move like a real person, he filmed his brother sprinting and jumping in the yard and traced the footage painstakingly by hand, frame by frame, onto a machine that barely had the memory to hold it. The journals are full of doubt and money trouble and the same jump animated over and over until it finally felt right. What moves me is the patience of it. One person bent over a vision only he could fully see, refusing to ship until it matched what was in his head.

He was the builder and the verifier in the same brain. Verification is the word everyone reaches for with AI now. Can you trust what the model produced, and can you check that it’s right. For code, and for anything chasing AGI, that’s the hard part, because the standard sits outside you. Someone else, or some test, decides whether the answer is correct.

Build for yourself and that problem changes shape. The standard isn’t out in the world. It’s your own taste, and you carry it with you. Verification stops being an audit and becomes a feeling: do I like this, is this what I wanted.

I felt this on a small thing recently. I built an app recently that turns my own handwriting into a font, and the only test that mattered was whether the result looked like my hand. Code at least has tests; another engineer can run it and tell you it works. Handwriting that’s meant to be yours has no such check. I would generate a set of characters, look at them, and know in a second whether it was me or a stranger doing an impression of me. I was the only person who could verify it, because the standard was my own hand and I was the only one holding it.

Mechner could answer the same kind of question about every frame, because he was the only person the game had to satisfy. You are the builder, the verifier, and the audience, all the same person, and the gap between wanting something and checking it shrinks to nothing.

That fusion used to be rare and expensive. Holding a whole vision yourself meant doing all the work yourself, which is why the game took years, and why most things got built by committee instead. AI takes the cost out. One person can hold the whole vision again, because the labor that used to need a studio now answers to a single pair of eyes.

Taste isn’t one ladder with the masters at the top. It’s personal, all the way down. Each of us has a private sense of what we wish existed, and for most of history that sense stayed locked up, because the distance between wanting a thing and making it was too far to walk alone. That distance is collapsing for everyone at once.

Some private visions will turn out to be everyone’s. A personal itch will match what millions of people quietly wanted, and the weekend tool becomes a company. That was always rare and stays rare. The one odd person whose taste happens to be the world’s is lucky in a way you can’t plan for.

It doesn’t matter. The hit was never the point. Most of what gets built now with AI may only be for one person, and that’s fine. A tool that fits your own hand and does the single thing no company would bother to make, because the market is exactly you, is worth building on its own. The making is the reward. AI didn’t give everyone Rubin’s ear for what a crowd will love. It gave everyone his stance: make the thing you want, for yourself, and leave the audience for later. For the first time, almost anyone can take that stance and actually finish.

So build what you want. The world may come, or it may not. Either way, you will have made something for the one person who was always worth building for.