In 2013 I ran a startup, and it took six months to take our first payment. Three months to get a WorldPay basket set up on the site, then another three before the remittance came through and the money actually landed. Half a year, end to end, for the most basic commercial act a business can perform. And here is the strange part: nobody around me thought that was slow. That was simply the speed at which things got made.
I have been thinking about that six months a lot lately, because the same job now takes an afternoon. The collapse in the time between "I have an idea" and "the idea exists, live, on the internet" is the biggest change to creative work in my professional lifetime, and it has quietly broken the machine I spent fifteen years working inside.
That machine was the agency conveyor belt. I worked as a creative strategist in brand and culture, and the deal was always the same: the strategist has the idea, the creative team makes the idea, production finishes the idea, the client pays for the idea. Everyone touches it once and passes it along. It was a sensible system for as long as making things was expensive and slow, because it concentrated the expensive part, the making, in the hands of specialists and rationed everyone else's access to them. My section of the belt came first. Which means I spent a decade and a half having ideas and then handing them to other people to find out whether they were any good.
The canonical case is Compare the Market's meerkat: a fictional character that became one of the most recognisable brand properties in Britain, spawning toys, books and over a decade of campaigns. Strip it back and the meerkat began as the kind of thought a strategist has in a Tuesday afternoon brainstorm. The thought cost nothing. Everything after the thought required an agency, a production budget and years of someone else's making. The system worked, but look at where the leverage sat: not with whoever had the idea, with whoever could manufacture it.
Last week, on a call with two founders I am partnering with, I heard myself compress what has changed into one sentence: you can import your taste and then have them build it. The "them" is AI tooling. The models are the creative department now: they write, they design, they code, they speak. What they do not have is a point of view about what is worth making, what good looks like, or when to say no. That part does not import from anywhere. That part is the job.
Here is what that looks like in practice rather than in theory. I am currently running an AI football pundit called the Gaffer: a fictional character with a comic-book face, a voice generated by ElevenLabs, and a World Cup prediction game with a couple of hundred players in it. I write his scripts the way a sitcom writer writes a character, three sentences of absurd anecdote adapted to whether his prediction won, lost or drew, and at full time a voice note is generated automatically and lands on the site. The conveyor belt version of this idea needs a voice actor, a recording studio, a developer and a producer. The collapsed version needs taste and a laptop.
Before that I built Buggy Smart, a map of London venues that can actually fit a pram, because I have a two-year-old and the dataset did not exist. An AI agent made over 10,000 phone calls, each one asking the same polite question in a synthetic voice called Poppy, and as I write the live map holds over a thousand verified spots: 1,273 at the last count. The conveyor belt version of that project is a call centre and a budget I do not have. The collapsed version cost less than a thousand pounds.
I have never written production code by hand. For most of my career that sentence was a confession. It is turning into a credential, because it describes exactly the shape of the new role: someone who can hold a standard in their head, brief precisely, judge output ruthlessly, and ship. The fashionable word for this is taste, and I used to think it was a soft answer. I no longer do. When the making is abundant, the deciding is the scarce thing.
Two parts of the old belt did not collapse, and they matter. The first is the idea itself: the models will build a bad idea just as quickly as a good one, faster in fact, because the bad idea meets less resistance in your own head. The second is distribution. More people are making things than at any point in history, and roughly the same number are making money from them. The app stores fill up faster than anyone's ability to discover what is in them. So the scarce ends of the belt are the two I now spend my time on: taste at the front, distribution at the back. The expensive middle, the manufacturing, is what fell away.
That rearrangement changes who owns things, which is the part I find most interesting. Under the old system, the people who manufactured the idea often ended up owning the most valuable piece of it, because making was where the leverage was. The meerkat made fortunes for more parties than the one who first thought of a meerkat. When a strategist can build the thing, the strategist can own the thing: the character, the product, the audience, the IP. That has never been true before, not at this cost, not at this speed.
I do not think the conveyor belt disappears. Big ideas at big scale will still need many hands, and the best creative companies will reorganise around judgement rather than manufacture. But for one person with a point of view, the belt is now optional. I spent fifteen years passing my ideas to the left and waiting. The wait is over, and honestly, the queue behind me should empty fast: every strategist I rate has a drawer full of ideas that only ever needed a maker. They have one now.
