The hat nobody can hand off

The launch film for Keyline is sixteen seconds long. For the soundtrack I did the clever thing: I wrote a small script that generated the score as a function of time, so the music was produced by the same logic as everything else in the product. It ran. The structure was correct. Every beat landed where the math said it should.

Then I played it, and it was wrong. Not broken, just lifeless. I spent about ten minutes on Pixabay, licensed a track, dropped it in, and the film worked.

That is the whole build in one story. Keyline went live at app.keyline.studio with real payments running, and I made it wearing every hat. Engineering, product, QA, support, the person who notices the checkout is broken on a Sunday. What surprised me was not how many hats there were. It was how many of them got lighter once I had agents working from a written brief, and how one of them did not get lighter at all.

Writing the score: transferred. Judging the score: mine, and only mine, and it took ten minutes.

Here is the uncomfortable part for those of us who lead design teams. Most of what I was proud of doing for twenty years was production. Careful, skilled, hard-won production. And production is exactly the part that moved. The part that stayed is the part I never learned to describe, never put in a job description, and never really tested for when I hired people.

Because look at how we hire. We ask for a portfolio, which is a record of output. We ask the candidate to walk us through a project, and they narrate the output. We admire the case study, which is output with a story wrapped around it. Then we hire someone to do a job where output is the cheap part, and we act surprised when they need a fully built environment to be effective.

In the second playbook I published here, on hiring your first designer, the practical exercise is scored on the questions the candidate asks, not on what they produce. One of the red flags in the scorecard is a candidate who shows only polished finals and cannot name a single decision they reversed. I wrote that line about small teams in engineering led companies. It reads differently now. It applies to every design hire I would make today.

So, the one thing.

In the next interview you run, cut the question "walk me through this project." Replace it with this: show me something you decided not to ship, and tell me how you knew. Then judge the answer, not the artifact. Listen for whether they can name what was wrong in words another person could use. "It felt off" is a start; "the tone promised a level of service the product could not deliver" is the job.

If they cannot produce a rejection at all, you are looking at a producer. Producers are still valuable and I would still hire some. Just know which one you are buying, because production is the part that transferred.

I am still working this out in my own setup. My brief gets longer every week, and it grows in exactly the places where I had been carrying taste in my head instead of writing it down. Every vague line is a decision I will have to make again by hand at midnight.

So the recap: production moved, judgment did not, and our hiring process is still aimed at the thing that moved. One question, asked early, points it back at the right target.

What was the last piece of work your team killed, and can anyone on it explain why in words you could hand to a stranger?