School's out for summer but AI is still in session — and it got faster again this week, with new agentic models that spin up sub-agents and build full apps from a single prompt.
This week: why AI listing images keep going wrong, how to combine trained VAs with AI, a new ad platform with a first-mover window, and a TikTok Shop stat worth sitting with.
AI Images Are Easy. High-Converting Images Are Not.
Here's the problem with faster AI: when AI gets faster, bad workflows get faster too. Nowhere is that more obvious than Amazon listing images.
On this week's OCEA alumni call, a member shared a real one. She used AI to create images for one product and it worked beautifully — smooth workflow, good output, almost no edits. Then she tried the same approach on table tennis balls.
It fell apart. The product came in multiple pack sizes: a 6-pack, 12-pack, 72-pack, and a bag of 140. The AI made the sizing look wrong. A smaller pack looked huge next to the bigger bag.
That's not a small issue. That's a product truth issue. If the image looks good but misrepresents the product, you don't have a high-converting image. You have AI slop.
And on Amazon, AI slop hurts you. It confuses the buyer, sets wrong expectations, and can lead to complaints, returns, or compliance problems when the visuals are off.
So the lesson wasn't "prompt harder." It was "turn the working workflow into a repeatable system." That means clear product references, accurate dimensions, correct colors, competitor context, buyer pain points, compliance rules, image sequence logic, and human review before upload.
Here's the line I keep coming back to: AI does not create high-converting images. Systems do. AI is the speed layer. The human is still the judgment layer.
That's why I brought Kamal Singh from Listing Optimizer AI onto this week's live training. Kamal has been making Amazon listing images for years through AMZ One Step, and he's seen the old agency model from the inside.
Instead of starting with a blank prompt, his tool starts with the listing. It scans the product, looks at competitors, analyzes reviews, finds buyer pain points, and builds image concepts — main images, secondary images, and A+ content.
It can even simulate how your main image looks next to competitors in the search results. That part matters. Your main image doesn't live in isolation. It lives in the SERP, side by side with every other product. If a competitor's image is sharper, they get the click before the buyer ever sees your listing. Image work isn't just design. It's CTR, CVR, and eventually rank.
We tested it live — not with demo products, but on 3 real attendees' listings. One of the strongest examples was A+ content generated in German. Was it perfect? No. Some colors and angles still needed a human eye. And that's the point. The workflow got the seller most of the way there in minutes.
That's the 80/20. AI doesn't replace judgment. But it can compress the messy first 80 to 90% of the work — research, concepts, drafts, variations, image copy, A+ layout. Then a human reviews, edits, tests, and approves.
We're not moving from "human designer" to "AI designer." We're moving from random creative work to repeatable image systems.
TAKEAWAY: If listing images are a bottleneck, don't try to rebuild your whole catalog with AI this week. Pick one important ASIN, run the full workflow, and judge the main image, secondary images, and A+ ideas against one question: is this getting me closer to a repeatable image system?
Watch the replay with Kamal here: https://youtu.be/RSBxN-Hwl0U
Run it on one ASIN and see the main-image options it gives you. 80/20 readers get 20% off monthly with code GARY20: https://listingoptimization.ai/
That last point — human review before upload — is exactly why tools alone aren't enough. Someone still has to check accuracy, claims, sizing, color, and compliance before anything goes live. Our friends at VAA Philippines put together a free playbook on exactly that.
Everyone is talking about AI right now. But one question keeps coming up from Amazon sellers: How do I actually use AI in a way that improves execution?
AI can help with PPC analysis, creative ideas, listings, and automation. But after the initial excitement, most sellers hit the same wall: AI can surface opportunities, but someone still needs to decide what actually matters.
That's why the team at VAA Philippines combines trained Amazon VAs + AI tools to help sellers move faster across PPC, creative testing, and day-to-day execution — while keeping human review at the center.
For example: AI can analyze PPC data faster, but trained VAs still review performance, prioritize actions, and validate changes before they go live. The same applies to creative optimization — AI explores ideas and shopper insights, while trained VAs identify what actually makes sense for the brand and business goals.
It's not about more AI hype. It's about using AI to support smarter execution.
TAKEAWAY: If your team is stretched thin, the fix usually isn't another tool — it's a trained human who can run the tool and approve the output. Grab their free playbook and steal the workflow.
Anytime there's a new ad platform, the first movers get the huge advantage while their competitors haven't caught on yet. Prices are low and there's maximum land-grab opportunity. That's exactly where ChatGPT ads are right now.
Here's what happened. OpenAI started testing ads inside ChatGPT on February 9, 2026 — labeled "Sponsored," shown below the answer, for logged-in Free and Go users in the US. (Plus, Pro, Business, Enterprise, and Education tiers don't see ads.) OpenAI says the ads don't change the answers.
Then on May 5, 2026, OpenAI opened a self-serve Ads Manager at ads.openai.com. This is the part that matters for sellers: the old $50,000 minimum spend is gone. There's no minimum now. Approval takes about 2 days, and you can launch with whatever budget you want to test.
The system runs on an auction with CPC (cost-per-click) bidding. OpenAI's recommended starting max bid is $3 to $5 per click. Eligible categories so far include household and consumer goods, local services, travel and entertainment, and digital products and education — with more expected over time. (Source: OpenAI.)
One more for the readers in my world: OpenAI said it's expanding the ads pilot to Japan, the UK, Mexico, Brazil, and South Korea.
A couple of people in our private cohort group have already been testing it. Have you tested it out yet? It does require an approval process, and I recommend you give it a shot if you haven't already — before your competition catches on.
TAKEAWAY: If you sell household or consumer goods, go apply at ads.openai.com this week. Early ad platforms reward first movers with cheaper clicks and less competition. Worst case, you learn the system 6 months before everyone else. Just confirm your category is eligible first — the beta is still limited.
See if your category qualifies and start the approval here: https://ads.openai.com
Marketplace Pulse reported on June 11, 2026 that the top 1% of tracked US TikTok Shop sellers drive about 60% of GMV (gross merchandise value — total sales on the platform).
That's worth sitting with. A new channel looks democratized and wide open. But the sales still concentrate around a small group of strong operators.
TAKEAWAY: New channels look like open fields, but the winners are operators who run them hard — consistent content, consistent testing, consistent execution. If you're going to test TikTok Shop, commit to it or skip it. Half-effort gets you nothing here.
Watch the Replay: Kamal's Live Listing Image Training
If listing images are your bottleneck, watch this week's training with Kamal Singh before you start your Prime Day image prep. On the replay, you'll see:
How the workflow starts from an ASIN
How competitor research feeds image strategy
How review data turns into image concepts
How main image variations are created
How secondary images and A+ content get generated
Where AI still needs a human eye
How live attendees reacted when their listings were generated on the spot
Watch the replay here: https://youtu.be/RSBxN-Hwl0U
Then run it on one ASIN. 80/20 readers get 20% off monthly with code GARY20: https://listingoptimization.ai/
In case you missed it:
See you next week,
Gary
P.S. — The pattern across all of this is the same: AI is the multiplier, not the strategy. The sellers who win this year are the ones who build the system, then point AI at it. Make sure you're multiplying the right thing.
P.P.S. — None of this works without a human in the loop. AI moves fast, but someone still has to check the work before it goes live. That's the whole idea behind VAA Philippines' free AI + VA Playbook — how trained VAs keep AI execution accurate across PPC, creative, and listings. Grab it here