A Fish Auctioneer and a Language Model Do the Same Job

Nobody puts a fish auctioneer and a language model in the same sentence. One smells of salt and diesel. The other lives in a warehouse in Virginia that hums like a fridge the size of a stadium. But watch both at work and the resemblance is rude.

The auctioneer stands over a crate of tilapia at 5 a.m. He looks, he guesses, he talks fast, he sets a price, he moves the crowd. He is a prediction machine wearing gumboots. So is the model. Both take messy input and spit out a confident answer before you finish blinking. Both are usually right and occasionally spectacularly wrong. Neither will apologize.

So let us go to the market at dawn and follow the money. Not the hype. The money. Because that is where you can actually see what AI is doing to work across the continent and across the diaspora.

The dawn shift, before the tools arrived

A fish market at first light is a machine made of people. The boats come in. The porters run. The cold-room man guards the ice like it is gold, because at the coast in January, it basically is. The auctioneer reads the crowd. The mama at the frying station already knows how many customers will pass by 9 a.m. The record keeper — usually somebody's cousin with a exercise book and a pen chewed at the end — writes it all down badly.

Every one of these people is doing information work. Pricing. Forecasting. Matching supply to hunger. They just never called it that, because nobody in a gumboot calls it "data."

Then a phone shows up with a model inside it. And the quiet reshuffling begins.

Follow the money: who gets promoted

Here is the thing the robot-panic crowd keeps missing. AI does not walk into the market and fire everyone. It walks in and quietly promotes some people while demoting others. The trick is knowing which chair you are sitting in.

Start with the cousin and his exercise book. His job was to remember and to write. A model does both, faster, and it does not lose the book in the rain. That role thins out. Nobody announces it. One day the exercise book is just not there anymore, and the cousin is either driving deliveries or he has learned to argue with the app when it is wrong. Which, it turns out, is a real skill.

Now look at the auctioneer. You would think the prediction machine threatens the prediction man. It does not. It hands him a second brain. He still reads the crowd — the model cannot smell desperation on a buyer who needs fish for a wedding today. But now he checks yesterday's prices across three markets before he opens his mouth. His guess gets sharper. His leverage goes up. He was already the most valuable person in the yard. AI just widened the gap between him and everyone else.

AI rarely replaces a worker. It replaces a task, then quietly re-sorts the people who were doing it.

That is the pattern, and it is the same in Nairobi as it is in a diaspora kitchen in Manchester or a small accounting shop in Atlanta run by two sisters from Kisumu. The person whose whole value was "I remember things and I type them" is under pressure. The person whose value was judgment now has judgment plus a very fast, very tireless intern who never asks for lunch.

The diaspora angle nobody prices in

There is a second market at dawn, and it runs on WhatsApp. It is the diaspora coordination economy: the nurse in Leeds who also runs a small import side-hustle, the Uber driver in Minneapolis who does everyone's tax paperwork, the auntie who is quietly the operations manager for three businesses back home while raising kids abroad.

These people have always done two jobs at once, separated by six time zones and a bad connection. Their scarce resource was never money. It was attention. There are only so many hours before the night shift.

This is exactly the crack AI slides into. Not the flashy stuff. The boring stuff. Drafting the supplier message in three languages. Reading a contract at midnight so you do not have to pay a lawyer for the obvious parts. Turning a voice note from your mother — who will never, ever type — into a proper order. The diaspora worker was already an agent, in the technical sense: a human running background processes across continents. AI just gave that human more threads.

Follow the money here and it is subtle. The savings do not show up as a paycheck. They show up as hours reclaimed, as a side business that finally stops leaking, as one less specialist you had to hire. That is real leverage. It also never appears in any "AI creates X jobs" headline, because it is invisible. It is a tired person getting ninety minutes of their life back on a Tuesday.

Who loses leverage, honestly

Let me not sell you sunshine. Somebody at the market loses.

It is the middleman whose only product was access. The fixer who knew the price because you did not. The broker whose entire business was standing between the fish and the buyer and charging a toll for information that is now free on a phone. When knowledge stops being scarce, the tax you could charge on knowing things collapses.

That is not a tragedy or a triumph. It is a repricing. It has happened before — to travel agents, to the guy who used to know which office to bribe, to anyone whose moat was a fact everyone can now look up. AI just does it faster and to more people at once.

The uncomfortable part: some of those middlemen were feeding families. "The market corrects" is easy to say when it is not your toll booth. So the honest builder does not celebrate the loss. He asks the better question — what does that person know that the model still cannot? Usually it is trust, relationships, the ability to walk into a room. Those do not compress into a chatbot. The people who survive the repricing are the ones who move up the stack from information to trust.

A small lab note

This is roughly why, in the Ni Biashara lab, we keep circling private, on-your-own-device AI — a notebook that remembers your business without shipping every supplier price and family voice note to somebody else's warehouse. If the whole game is leverage, then the model that lives close to you, under your lock, is worth more than the clever one that lives in a stranger's fridge. Not paranoia. Just keeping your own books.

What to do now

You do not need a strategy deck. You need one honest hour with a pen.

• Write down the three tasks you do that are pure memory-and-typing. Those are the ones the intern should take first. Give them away without ceremony.
• Find the one task that is pure judgment — the thing only you can smell. Protect it. Sharpen it. That is your auctioneer skill, and it is where your pay is going to come from.
• If your whole value is standing between two people and charging a toll, start building the thing a model cannot copy: the relationship, the reputation, the room you can walk into.

The market at dawn already knows how this ends. The fish still have to be weighed. Somebody still has to read the crowd. AI did not empty the yard. It handed out sharper knives and quietly rearranged who gets to hold them.

Pick up a knife. Just check whose hand it sharpens.