Rick Rubin was asked a simple question on The Diary of a CEO: if an AI made ten million songs, would one of them be a hit?
He didn't argue. It could make ten thousand tracks, he said, and you might like one or two. Then he described how hip-hop records got made in the eighties. Producers went crate digging: hours in used record stores, listening to whole albums to find one short drum break worth building a new song around.
The records were never scarce. The ear was.
That is the clearest description I have heard of where small businesses stand with AI right now. Production just became nearly free. Judgment did not.
The machine has no point of view
Rubin's case is plain. AI is the collected ideas that already exist. Ask it the same question on three different days and you may get three different answers, because it has no fixed way of seeing. It can give you fifty versions of your intake email, your homepage, your proposal. It cannot tell you which one sounds like the firm your clients trust, because it has never sat across the desk from them.
You have. That is the part of your business nothing replaces.
Before you hand any task to an AI tool, write one sentence describing what good looks like. If you cannot write it, the tool cannot hit it.
Listen before you build
It is also why I have long respected Rubin. He is not the stereotypical producer. By his own account, he barely plays an instrument and does not know how to run a soundboard. His method is closer to a philosophy than a technique. He listens, and he asks questions until the artist can hear themselves.
He described a pattern on the show that I recognized right away. Artists come to him, tell him exactly what they want to make, then ask him what they should do. Most people, he said, are not used to being really heard. His job is mostly removing the roadblocks they have put in their own way.
Owners often do a version of this. They describe the problem in detail, then ask which software to buy. So when I run an assessment or an audit, I try to ask Rubin's kind of questions first. How does this part of the week feel? What does that show you? Where are you really on this? The answer is usually already in the room. The tools come after.
Pick one process that bothers you and ask yourself those three questions before you shop for anything.
The brief is the work
Rubin compares prompting to how Alfred Hitchcock worked. Hitchcock described every frame of a film before a camera rolled, and artists drew what he described. In Rubin's words:
The prompt is where the creativity is.
Anyone who has run IT projects knows the business version. Projects rarely fail at the tool. They fail at the requirements: nobody wrote down what the thing was supposed to do, for whom, and what counted as done. AI makes this worse: a vague request now gets you a confident, polished, wrong answer in eight seconds.
Treat your first AI project like a hire. Write the job description first: the task, the inputs, the output, and who checks the work.
Run the pizza test
Rubin explains his confidence with food. Hand someone two slices of pizza and ask which is better. They will not predict what other people would like. They will tell you which one tastes better to them.
Most owners evaluate technology the opposite way. They buy what a competitor bought, what the loudest vendor pitched, or what a webinar said everyone is using. That is guessing about someone else's pizza.
Run two options side by side, on your real work, for a week. Keep the one you would actually open on a Tuesday. Your preference is data.
Don't automate the average
AI pulls toward the average by design. It is built from what already exists, so its first answer is usually the most expected one. Use it to produce what everyone else produces and you will sound like everyone else, faster.
The owners who get real value use it the other way: to clear the busywork, so they have more time for the parts only they can do. The judgment call, the client conversation, the thing the firm is known for.
List the three things clients specifically hire you for. Automate around them, never through them.
The ear is the asset
The crate diggers did not own the record stores. They owned the ear.
Every business now has the same infinite record store. Your competitors can generate exactly what you can. What you bring is the ability to hear the one part worth building on, and the nerve to choose it.
That skill was always yours. It just became the whole job.
Part 2, in the next edition: the harder half of what Rubin said. An ear only works if you are free to use it, and most owners spend their days answering what is demanded of them. We will talk about getting out from under that long enough to find the work only you would make.