The War Between Brand and Performance Is Over. Both Agencies Lost. AI Won.
I published this in October 2025, the week OpenAI put apps inside ChatGPT. Updated May 2026. I've left the argument where it stood and updated what's happened since. The part I got right and the part I got wrong are both more useful now than the day I wrote them.
In October 2025, OpenAI put apps inside ChatGPT and shipped AgentKit, the kit that let anyone stand up a working agent in an afternoon. Canva and Spotify stopped being places you went and became things you called for mid-sentence. I looked at that and made a bet: the war between brand and performance was over, because AI had folded distribution, creativity, and conversion into a single moment, and you needed no media buy to reach the people inside it.
I was right that the war was over. I was wrong about where it would end. I thought it would end at the purchase, brand and buy collapsing into one tap inside the chat. OpenAI thought so too. They switched on Instant Checkout in February 2026 and pulled most of it back by spring, fewer than three dozen merchants ever live. Buying the thing inside the conversation turned out to be the hard part and the part that mattered least. The split was dead, that much was easy to see. Where the functions would fuse instead, I called wrong.
Because while everyone watched the checkout button, the fusion was happening somewhere else, and it was bigger than a purchase. Performance went first. You no longer buy media. You hand the platform a budget and an objective and it picks the audience, the placement, the creative, the bid. Google took its AI buying tools out of beta and began retiring the ones media planners spent careers learning. On the other side, one brief now produces the film, the fifteen-second cut, the landing page, and the try-on before lunch. Brand, performance, product, experience. Four departments, four budgets, four elevator banks, and a machine that treats them as one job.
So the functions are merging. Everyone in the room already knows that, and saying it out loud in 2026 earns nothing. What nobody says is that merging them is the easy sentence and the hard job. A machine can generate the four assets. It cannot decide which idea is worth four assets, or hold a brand steady while performance drags it toward whatever converts this week, or know when the elegant system is the wrong answer because the elegant system has no point of view. That judgment, the kind that runs across all four functions and answers to none of them alone, is the scarce thing now. It does not come from the model. I have spent a career being the person in the room who carries it, on work that sold out the product, signed up the millions, and still looked like something.
I was right that the war was over. I was wrong about where it would end.
The question stopped being whether you are a brand agency or a performance agency. It is whether you are still organized as four departments or as one practice. The smartest people in this business have been circling the same conclusion from different sides for years. The old dichotomies are dying. The industry spent decades arguing about the line between media and creative, and AI erased the argument while we were still having it.
And yet most of the conversation is still about the wrong thing. While the trade press debates whether AI will kill agencies, finance chiefs are taking marketing departments apart one false choice at a time. Brand or performance. Art or algorithm. The product team, or the people who name the product. The sixty-second film nobody watches past the second second, or the banner that follows you to your grave. Every one of those choices was always a fiction, and the fiction is getting expensive.
Patagonia built a billion-dollar brand by running a Black Friday ad that told people not to buy the jacket. In the nine months after, sales climbed about thirty percent. That was never brand instead of performance. It was conviction that happened to convert, which is a thing the binary has no category for. Patagonia did not need ChatGPT to reach the planet. It needed something to say. But picture that same conviction built into an agent that helps you repair, resole, recycle, and once in a while buy. The same point of view, awake at all hours. That is not the end of brand. That is brand that never sleeps.
The old way of charging is already gone. You cannot bill hours for work a model finishes in seconds, and you cannot bill for execution when anyone can execute. So the fee has to move to the thing that stayed scarce: knowing which agent the world actually needs, which idea is worth building, which problem is worth solving at all. Licensing for the systems you build. Outcomes tied to what they produce. Revenue from products that live inside a client’s business instead of deliverables they file and forget.
When anyone can build an agent in ten minutes, agencies can’t charge for execution anymore. We have to charge for knowing which agent the world actually needs.
When the model shifts from selling time to selling what the time produces, creative finally gets permission to be brave instead of safe. This is not theory for me. The blueprint exists, the technology works, and the hard part was never the technology. It is the translation, moving between the creative, the corporate, and the consulting worlds and carrying an idea intact through all three. That is the job now. Not maintaining. Building.
We are watching the grand campaign die. That beautiful expensive thing we used to unveil like a debutante, all promise and no follow-through. On the other side, the performance machine retargets you into madness with ads for the thing you already bought, forever, like an NPC with one line of dialogue. Both camps are still fighting a war that ended before they noticed.
The work now is not doing the old marketing faster. It is building how marketing actually works when intelligence is cheap. Systems that compound instead of campaigns that launch and vanish. Products that become part of a client’s infrastructure. The question is not whether to change but how fast, and the firms moving are already building the capability next to the creative department, not in a lab down the hall.
When anyone can generate the asset, the value is not in the asset. It is in knowing which one is worth making. When any brand can reach anyone, the problem is not reach. It is resonance. When the machine makes infinite variations, the scarce thing is not content. It is conviction.
When AI can generate infinite variations, the scarcity isn’t content. It’s conviction.
The opportunity is not survival. It is to finally be the thing clients have always needed and rarely gotten: one partner that does not split itself into warring tribes. The brand poets and the performance quants and the product team that has never once sat with either. The finance chief looks at the attribution on the banner and nods. The marketing chief looks at the Lion on the shelf and smiles. Neither one is looking at the person who just wants something that works and moves them at the same time. Apparently still a radical request.
Remember when digital was its own department? We laugh now. The digital team in the basement, the traditional team in the penthouse, separate elevator banks and all. Then we rebuilt the same dysfunction at scale. Brand teams who have never spoken to a customer. Performance teams who think craft is a kind of cheese. Product teams shipping features nobody asked a human about. The beautiful work no one sees, the ugly work everyone sees, and the slow collapse of an industry that forgot its only job was to make things people want.
Performance teams who think craft is a kind of cheese. The beautiful work no one sees, the ugly work everyone sees.
The firms that win will not be the ones defending last year’s margins. They will be the ones building next year’s model, turning creative force into technology that compounds, charging for outcomes instead of outputs, turning the one-time campaign into something that earns every month. This is where creative and corporate and consulting finally converge, and where the people who can speak all three become the ones worth hiring. Builders over maintainers.
The future does not belong to agencies or to consultancies. It belongs to whoever builds the new model first, with the taste of the best brand shop and the precision of the best performance shop and the spine to hold both. Agents that write poetry while they optimize. Algorithms with a point of view, not just a conversion rate.
Algorithms with a point of view, not just a conversion rate.
Maybe that is the real shift. Not the death of agencies but the death of the false choice. Not brand or performance but brand through performance. Not beauty or utility but beauty as utility. Not the campaign that launches and disappears but the system that learns and stays. Not product on one floor and story on another but one thing, made by one room.
When Patagonia told people not to buy the jacket, it was not choosing between brand and performance and product and experience. It was choosing a point of view, which, awkwardly for everyone holding a media budget, beats all four. The best growth hack turns out to be caring.
Is this the end of agencies as we knew them? Of course. But endings, like false choices, are mostly just beginnings somebody was brave enough to build. The ones who make it through will not be the ones picking a side in a war that is already over. They will be the ones who saw that the war was always a fiction, a convenient excuse for mediocrity dressed up as specialization, and who learned to hold all of it at once. Brand, performance, product, experience. One practice. One point of view, carried across the whole thing by someone who can actually carry it.
The revolution is not coming. It is here. And it needs builders.