The Last Thing to Scale

Everyone is talking about what the machines can do now. Generate a face. Clone a voice. Build a whole performance out of nothing but a few photographs and some compute. The tools got good, and then they got cheap, and a lot of people looked at that and decided the game was about the tools.

It is not. It never was.

Because once the technology is everywhere, the technology stops being the advantage. Models improve. Platforms multiply. Engineering can be bought by the hour. If your edge is the tool, your edge has a shelf life measured in months, and there is always a better-funded room somewhere building the next one.

The hard questions were never the technical ones.

Who has the right to authorize this face? Is the consent real, or was it clicked through by someone who did not understand what they were agreeing to? What exactly is the buyer allowed to do with it? Can that permission be taken back? Who gets paid, and when? And when something goes wrong, and something always goes wrong, who is responsible?

None of those are engineering problems. Every one of them is a trust problem. And the company that can answer them, reliably, in a way that the talent believes and the buyer believes and a court would believe, becomes more important than the company generating the image. The picture is the easy part. The permission behind it is the whole business.

Here is the thing most of the decks miss, the ones written in glass towers by people who have never sat in the room. Trust does not scale the way code scales. You cannot copy and paste it. You cannot raise a round and buy it. Trust is not a protocol. It is cultural, and it is carried by people.

A startup in London can build the finest platform in the world and still be a stranger at the gate in Lagos. In the markets I come from, trust does not arrive by email. It is carried in by someone the room already believes. The person who vouches matters more than the terms of the contract. Nobody gets paid alone. Money travels as rumor. A yes from the right person opens a door that a thousand cold emails will never touch.

I watched this happen in my own life this month, and it did not go the way the story is usually told. For a year I have been building with these tools and telling anyone who would listen that they should learn them too. The people best placed to judge the work, the ones who could actually see what I was doing, mostly told me it was not that deep. Then a man who could not have named a single tool I used sent me money for it, without being chased and without being asked twice.

He is my uncle, not by blood, which where I am from often means the bond is deeper rather than thinner. For seven years his was the door that opened. The house everyone gathered at, the connection to home for people a long way from it. He did not ask me what the technology was. He did not ask whether it was that deep. He had watched me for a long time and decided that whatever I was building, I would not embarrass him with it. That is not a purchase of software. That is a purchase of a person, and the software arrived with him.

The ones who could evaluate the work did not trust it. The one who could not evaluate it at all did. I did not have to go looking for the argument of this essay. It came and found me at my own family's table.

And this is exactly why the places the world keeps calling behind are actually the opening.

Africa does not have the formal rights machinery the West spent decades building around its agencies, its unions, its labels, its lawyers. From the outside that looks like a gap. From the inside it is a greenfield. There is no incumbent standing in the doorway. There is no widely adopted operating standard telling everyone how this is supposed to work. Which means the first person who can stand in that space and be believed gets to set the standard, before the regulators arrive and before the big agencies notice the continent exists.

That person is not the one with the best model. It is the one who can gather real trust and wire it to the world.

The badge, the verification, the mark that says this person is cleared and this consent is real, all of that is just the artifact. It is the receipt. The trust is the thing being sold. The paper is downstream of the person who could get the yes in the first place.

So the move was never to import trust into the market. You cannot. The move is to carry it. To be the bridge between the talent who has never once been paid or protected or recognized on the world stage, and the buyers who need someone they can actually believe. Local trust, global rails. The hand that both sides will shake.

And the best part is that it repeats. What you build in the first market is not a one-off deal. It is a method. You prove it where you are from, where you already have standing, and then you carry that method into the next room, and the next. The technology stays roughly the same everywhere. The trust has to be rebuilt in each place by someone who belongs there. That is not a weakness in the model. That is the moat. It is the reason this cannot simply be copied by whoever has the most money.

We spent a few years being told the future belonged to whoever had the biggest model. I think that was always half the story. The models are becoming a commodity, the way electricity became a commodity. What stays scarce is the thing that was scarce before any of this: the ability to make a promise that people believe, and to keep it.

In the age of the machine, the scarcest asset turned out to be the oldest one. Not the code. The handshake. The yes from someone the room believes.

That is the last thing to scale. And it is the whole game.