This morning's OCEA alumni call reminded me why this space is so exciting — and why it still needs judgment.
We're seeing real ecommerce workflows start to work with AI agents: research, listing workflows, reports, content, and repeatable tasks that used to sit in someone's head or spreadsheet.
But there's a catch.
The constraint is not just whether the AI can do the work.
The constraint is whether the workflow is safe, compliant, repeatable, and still valid 3 weeks from now.
Amazon TOS matters. Human review matters. And the tool that worked last month may behave differently today.
That's why I'm less interested in cool prompts and more interested in reusable AI workflows sellers can actually trust.
Here's what I mean.
One of the sellers in OCEA runs a large 8-figure ecommerce brand in Europe. The team is reworking product titles across about 500 products in 5 countries.
They handed an AI skill the company's own master file — current titles, bullets, descriptions, and keywords — and gave it one job: propose a new 75-character title and Item Highlights for every product. It produced a first serious draft across all 500.
But the first pass was not good. The AI got too technical, stuffing the titles with product specs most shoppers would never type into search.
So the operator corrected it. Keep the main buyer keyword in the title. Keep the language customer-facing. Push the technical detail down into the bullets and A+ Content, where it belongs. The next pass was much better.
The win wasn't that the agent understood 500 products. It's that it could draft across all 500, take the correction, and leave the operator in control of the standard.
AI prepares. The operator approves.
Here's the 80/20 of Ecom for this week:
In today's issue:
Why the real AI win is a workflow you can trust, not a clever prompt
The Amazon 75-character title change, and how to treat it like a catalog migration
China now owns Amazon's top 10,000 by seller count — but U.S. sellers keep the GMV
Why TikTok Shop is growing but is not an easy blue ocean
A Prime Day "Selling Fast" badge that grabbed attention, then vanished
Treat the Amazon title change like a catalog migration
Here's the change behind that project, and a lot of you will hit it.
Starting July 27, Amazon says product titles in every category except media need to be 75 characters or fewer, including spaces. It is adding another 125 searchable characters through Item Highlights — materials, use cases, comparison details.
Read the fine print: July 27 is the start of a gradual rollout, not a hard deadline. Titles still over the limit get updated over time using Amazon's own AI recommendations, and brand owners get a 14-day window to review, change, or approve before those changes go live.
For a large catalog, this is not a copywriting task. It is a migration. If I were running it, I would do what that seller did — keep the AI on a file you control, and approve every change before it goes live:
Export your current catalog and keep an untouched copy you can roll back to.
Start with your top-revenue and over-the-limit ASINs, not the whole catalog.
Generate proposed titles and Item Highlights in small batches.
Check keyword relevance, customer clarity, claims, and variation consistency.
Test a small set before you scale it across everything.
TAKEAWAY: Don't let Amazon's AI make the first call on your best listings. Start with your top ASINs, build the review step, and keep the original file so you can roll back. AI prepares the draft; you approve what goes live.
7 Figure Seller Summit 12 — July 28-29, online
This is exactly the work we'll get into at 7 Figure Seller Summit 12: putting AI agents to work on PPC and marketing, and on protecting profitability — with a human still approving anything that touches money, rankings, or account health.
It's 100% online.
July 28-29, 2026.
Marketplace Pulse looked at Amazon.com's top 10,000 sellers. Chinese sellers now make up 55.9% of that group, up from 42.5% in 2020. U.S. sellers fell from 53.7% to 40.5%.
But seller count is not economic power. Marketplace Pulse says U.S. sellers still generate 65.3% of the cohort's GMV, versus 28.6% for Chinese sellers — most of it from U.S. brands dominating the top 100. The lazy read is "China won Amazon." The real one: China is winning on breadth, while established U.S. brands still hold the scale at the top. (These are Marketplace Pulse's cohort estimates, not Amazon's totals.)
TAKEAWAY: Don't measure your position by ASIN count. If you're a U.S. brand, your edge is scale, brand strength, and margin — protect it. If you're competing on breadth, you're in the most crowded part of the market, so price and source accordingly.
TikTok Shop is growing. That doesn't make it easy.
Scott Needham of SmartScout used to see TikTok Shop as an easy blue ocean for young brands. After studying it and selling on it himself, he changed his mind.
His early data shows roughly 80% of unit sales going to Chinese sellers. Treat that as preliminary — SmartScout hadn't published the dataset when he shared it. But the direction is the point: womenswear looks strong, the platform is still building brands, and it's growing. Growth just doesn't mean easy entry. Chinese sellers bring supply-chain speed, price pressure, and a firehose of content.
TAKEAWAY: Treat TikTok Shop like a content-and-commerce system you're testing, not a lottery ticket. Set your category, margin floor, creator budget, and a testing window before you scale — and pull the plug if the numbers don't clear.
Charlie Banks spotted Native Pet adding a bold red "Selling Fast" badge right onto its Prime Day main images. Listing specialist Kamaljit Singh amplified it, because the badge sat directly above Amazon's own Prime Day deal treatment and used the same color family — two stacked urgency signals in the search results.
Then it disappeared. An official Sell on Amazon account said the images appeared to have been corrected and pointed to Amazon's image policy. There's no published evidence the badge sold a single extra unit, and I wouldn't copy it — it looks like an image-policy problem. The real lesson is underneath the tactic: brands are engineering search-page attention through visual hierarchy, packaging, color, and urgency.
TAKEAWAY: Test the attention principle without the policy risk. Use real packaging, compliant callouts, and clear visual hierarchy — not badges that mimic Amazon's own — before you touch your highest-traffic images.
The Vital Few — this week's links worth saving:
Talk soon,
Gary
P.S. If you want to see how sellers are putting AI agents to work on PPC, marketing, and profitability — with the operator still in control — get your limited-time free pass now.