AI Made the Making Cheap. Now Sell the Deciding.

The making got cheap. That is the part your team is feeling right now, and it stings, because the making is what most of them spent a decade getting good at. But the thing around the making, knowing what is worth building, having taste, defending a call, that part just got more valuable. Let me catch you up on what shifted and what to do about it before your next review.

The cheap part is the part you built your rates on

Here is the uncomfortable truth an anonymous designer named out loud: "the thing I've spent 15 years getting brilliant at, the making, the craft, the production, is becoming the cheap part." That is the shift in one line. Your billing model was probably built on hours of production. Those hours are worth less now.

This is not new, it just moved faster. As Claire McDivitt at Lazerian puts it, "Photography didn't replace painting. Adobe didn't replace designers." Each tool changed what was valuable. AI is another step, and it moved the value from your hands to your eyes.

So stop selling hours of output. Start selling the judgment behind what gets made. Brand designer Michael Bullamore said it plainly: "anyone can generate something that looks finished now, fewer people can tell if it's right." That gap is your new product.

Make the deciding visible, or nobody pays for it

The problem with selling judgment is that it is invisible. A client can see a logo. They cannot see the ten calls you made to get there. So you have to show the work you used to hide.

Start with case studies. David Sedgwick of StudioDBD says the shift begins in how you frame the work: name the problem, the decisions along the way, and the difference it made. Move from "I make things that look good" to "I solve problems." Ian Paget at Logo Geek tells every client straight out that they are not buying a logo, they are buying the thinking that leads to it.

For your team, this means one concrete habit. Have designers write down each decision and why, as they make it. Experience designer Nuria Quero does exactly this: "Every time I make a decision, I write down the reasoning." Six months of that, and a designer has a record of their own taste that no model can copy. That record is also what you show in a pitch.

The upkeep is the job nobody demos

The cheap prototype is a trap if you think it means the work is done. Dan Maccarone, a fractional product leader, watches this happen for a living. A model spins up a gorgeous prototype in an afternoon. Then "the upkeep is yours, forever, by hand." One of his designers spends real time fighting the model to keep a component library current. Tell it to update the thing, watch it not update the thing, tell it again.

AI does the doing. You still do the deciding, and the checking. Simon Willison put the line where it belongs: the one thing you cannot hand to the machine is checking that the thing actually works. Skip that and you do not have a working system, you have a hope.

The teams that survive make the invisible work legible. Maccarone points to evals, a suite that proves the product is a little more right this month than last. It is the closest the upkeep ever comes to a demo, and it is what you should be funding, not just the launch.

Regulation is judgment work you can charge for

If you want proof that judgment is the paid part, look at where the stakes are highest. In fintech, insurance, and healthcare, a clean screen is not enough. Andy Bhattacharyya argues the practitioners who thrive there stop treating compliance as the enemy and start treating it as a user need wearing different clothes. GDPR and HIPAA are the codified version of what users want: transparency, security, the right to understand.

The skill is pulling compliance upstream, into discovery, where it is cheap to accommodate. Wait, and you redesign screens three weeks before launch. Bhattacharyya also makes documentation part of the deliverable: a decision you cannot defend later is a decision you will have to undo.

A model cannot do this for you. Deciding what is worth fixing, and defending the call in front of legal, is exactly the judgment AI can't fake. That is billable, senior work, and it is a reason to keep experienced people on the team, not fewer.

The deep cut

Here is the thing that is easy to miss under all the repricing talk. Frota warns that the danger is not just cheaper output, it is that your people stop making decisions at all. If a designer hands everything to the model, they "literally forget what was done, because it didn't mean anything." No memory of the call means no ability to defend it, learn from it, or repeat it. And judgment is the one thing you are now selling.

So the practical payoff is not "use AI less." It is this: protect the moments where your team actually decides. Frota's line is blunt, "if we use tools without any criteria, we go from being designers to being operators." Operators are the cheap part. Build a rule that says which calls a person makes and can explain, and which the model drafts. That rule is what keeps your team's value from collapsing along with the cost of production.

Three questions for your team

1. Where are we still billing for hours of production instead of the judgment behind what gets made, and which client can we test a new frame on first?

2. Which decisions on our current projects can a designer explain and defend, and which ones did the model just make for us? If nobody can answer, that is the gap to close.

3. On our AI-assisted work, who owns the checking and the upkeep after the pretty prototype ships, and is that work funded like the launch or forgotten like an afterthought?