At an AI lunch-and-learn I gave for a tactical gear maker, someone asked what the bigger companies in their industry are doing with AI.
I think they expected something exotic. A secret model. A custom platform with a price tag they could never justify.
I told them about two sellers I know.
A month of work, in a day
The first is a 7-figure Amazon reseller. Their business lives or dies on one number: how many listings they can create.
A good listing is a heavy lift. Photos, keyword research, the title, the bullets, the infographics. Then a flat file with dozens of fields, half of which will never apply to them.
“Before, I would have a team of people and it would take one month to launch one product,” he told me.
He spent days going back and forth with Claude, building a dashboard that pulls in the product data and fills in the fields. The first product he pushed through it went live within five minutes, once a couple of field errors were fixed. The flat file, the worst part of the job, takes an hour or two.
I asked him: so you took a month of work with a team down to a day?
“At least.”
The launch numbers say the same thing. By the end of July 2026, they had run five rounds of product launches, more than in all of 2025. And 90% of those products were selling.
Then he did the same thing to advertising. He built an automation in Claude that writes the ad campaigns for each new product. It pulls keywords from the campaigns that are already winning. It learns from the ones he paused for overspending. It skips the brands that never convert. It knows a ranking campaign from a defensive one.
Then it stages the campaigns and waits. When the listing goes live, the campaigns go up to Seller Central, and he clicks Enable.
His goal is 10 product launches a day.
Less ad spend, more sales
The second is a founder who built an 8-figure, 85-person brand from nothing over 15 years.
In December 2025, he fired his ad agency. It was costing about $15,000 a month and, in his words, “really not doing that great of a job.”
He took the ads back himself, with Claude connected to his Amazon data through our MCP Server.
The agency had run 12 campaigns. He runs about 40, each built around one product format with its own keywords, so a shopper searching for one thing sees an ad for that thing.
Four months later, he was saving about $75,000 a month: $15,000 from the agency fee, and about $60,000 in ad spend, with daily spend down from $5,800 to $3,800. Cost per click on some of his ads had dropped from $2.50 to $0.90. Sales were up by a third. One product went from selling 400 to 500 units a month to 1,400.
“I feel like I feed Claude 10%, it does 80%, and then I watch the last 10%,” he told me.
It didn’t start that way. At first, the AI made changes on its own, and he stopped it. Since then, it tells him what it wants to do, and why, and he approves every change.
He uses the same AI beyond advertising, including to catch refund fraud.
He’d stumbled on one Amazon shopper while scanning his refunds. Over two years, her pattern grew. At first she ordered a single item and got refunded without sending it back. Then it happened again a month later. Eventually she was claiming orders never arrived before they could even reach her, even though Amazon was shipping them from its own warehouses. Her orders grew to three, four and five items at a time, about $250 each, and every one ended in a refund with nothing returned. By the time he caught it, she had taken $5,200.
Then he had Claude search his orders for other buyers with high refund rates shipping to the same address. It turned up two more accounts: three in all, used one after another, all going to her address.
What they have in common
Neither of them did anything exotic. No custom model, no enterprise platform. They use the same AI you can download.
Here’s what they did instead.
- They cleaned up their data first. The reseller’s breakthrough wasn’t the AI. It was a year-long project to build one source of truth for their product data, so the AI had something it could rely on. “The process to get there was the thing that unlocked going from a month to a week, down to a day,” he said.
- Someone owns it. Not a committee, not a pilot program. The reseller built his dashboard himself, one conversation with Claude at a time. The founder took the ads back himself.
- They do it every time, not once. Every new campaign builds on the last one. Every process that works gets written down as an instruction the AI reuses, so nobody explains it twice.
- A person still makes the call. The reseller’s team reviews every listing before it goes up, and he clicks Enable on every campaign. The founder approves every ad change.
Why smaller can be faster
Most companies are behind on this, including big ones. The ones ahead aren’t the ones with the biggest budgets. They’re the ones with clean data and a person whose job it is.
That’s an advantage for a smaller seller. At a big company, a change like this takes two quarters and a committee. At yours, it takes a conversation.
If you’re an agency
The agency in the second story ran 12 campaigns. The founder, who knew his products, ran 40.
That’s the bar for agencies: client-by-client depth that nobody could afford when every campaign was built by hand. AI makes that depth affordable for every client you have. The agencies that use it that way will keep their clients. The ones that don’t will be competing with their clients’ own AI.
The secret
The person who asked me expected a secret.
There isn’t one. The sellers ahead of you are using the same AI you can download.
They cleaned up their data, gave the job to someone, and kept at it until a month became a day.
