adlk.ioFor about fifteen years the hardest part of making software was making it. You had an idea on Monday and a working version sometime the next month, if you were fast and the idea was small. Building was the bottleneck. How we hire, organize teams and raise money was all shaped around that one fact.
That fact is gone. The cost of building has dropped to almost nothing and it's still dropping. I can describe a feature out loud and have a rough version of it running before the coffee is cold. So can you. So can the founder down the hall, and the sixteen year old who has never shipped anything.
When everyone can build, building is no longer what separates the products people keep from the products people forget. Something else takes that job. I think product people are the ones best equipped for it, and it isn't even close.
The expensive part was never the building
I spent a decade on one product. If you had asked me at any point what the hard part was, I wouldn't have said the code. The code was a lot of work, but it was work I understood. You sit down, you grind and it ships.
The hard part was the deciding. Who is this for. What earns their trust in the first 30 seconds. What do I cut. What do I refuse to build even though three loud users ask for it every week, and the enterprise inbox keeps promising money if I just add the one thing.
Around 80% of what I tried on Franz never shipped. Building it was never the problem. At some point I looked at the thing honestly and decided it was the wrong thing to put in front of people. Saying no was the most valuable work I did, and it had nothing to do with how fast I could type.
So when people tell me AI changes everything because now anyone can build anything, I agree with the words and not with the conclusion. Building was never the part that mattered most, and now it's barely a constraint at all. The expensive part is judgment, and that one hasn't got any cheaper.
A product is a story before it is code
I learned this slowly and now believe it completely: a product is a story before it's a single line of code.
Before I build anything I want to know the story. Who is this for, what does it promise them, why should they care, what is it not. When that story is clear and honest the product almost assembles itself around it, as every decision has something to check against. And at the end you're not bolting marketing onto a finished thing. You're telling the same story you've been telling yourself the whole way through, which is the only kind of selling that doesn't feel like selling.
Most good product people are storytellers, whether or not they would use that word. They hold the story in their head and keep the product honest to it.
AI doesn't give you that story. Ask it and it'll hand you a hundred stories, all quite reasonable, none of them yours. Picking the true one, the one you actually believe and can carry for years, isn't something you can generate. You have to mean it.
The tax that used to keep us honest
We're about to lose something here that was doing real work.
A lot of what made good products in the past was the tax of building things you weren't sure about. Building was so expensive that you couldn't afford to build on a hunch. Scarce resources at a company worked as a gatekeeper. Not always a good one, as good ideas died for no reason other than nobody having the capacity that quarter. But that gate forced something valuable. When you started a new path with your product, you had to validate it from A to Z first. You had to convince yourself, and usually other people holding the budget, that the thing you were about to build actually made sense.
This is why books like The Mom Test and Lean Startup still make 100% sense to me. Talk to your users before you build. Learn to ask in a way that doesn't fish for the answer you want. Validate the problem is real. Ship the smallest thing that tests the idea, then look at what actually happened. None of that was invented because it sounded elegant. It's the discipline the cost of building forced on you, written down.
The cost is gone now, and with it the thing that forced the discipline. But the discipline was never wrong. It was right for reasons that have nothing to do with how much building costs. So it has to move from something the budget enforced to something you choose to do on purpose, and chosen discipline is the kind most people skip quite happily. That validation work should stay part of the core muscle. It's more important now than it was, as nothing else is standing at the gate anymore.
Building is now a bonus, not the entry fee
I don't think technical skill stopped mattering. It didn't.
Being able to actually build is still worth a ton, maybe more visibly than before. I can look at what a model produces and feel when it's wrong somewhere underneath, or when the structure will hurt in six months and the easy path is a trap. That judgment makes AI-built products a lot better, and the people who have it will ship higher quality work than the people who don't.
What changed is where it sits. For fifteen years you couldn't be in the room without the ability to build. It was the entry fee. Now you can be in the room without it, and the people who have it bring a strong bonus instead of the one non-negotiable skill. Technical ability moved from the thing that lets you start to the thing that makes your work better once you're in.
The room is changing
I think this is good news, and not only because it happens to describe what I do.
For most of my career the product instinct was treated as the soft part of the work. The real work happened in the code, and the product person was the one with opinions about it. When building was the scarce and hard skill, that order made some sense.
It doesn't make sense anymore. When the cost of building falls to near zero, the value moves to the decisions around the building and to the story those decisions serve. It also moves to the validation nobody forces on you anymore. That's the product person's job. The people who spent years training it under real pressure, on real products with real users who left when they got it wrong, are about to be the most useful people in the room.
The tools changed. The hard part didn't.
If you're a founder working through exactly this shift, trying to figure out what to build when you can suddenly build anything, that's the work I do with teams. Let's talk.
