AI can generate 20 good Amazon images before lunch.
By Kamaljit Singh · September 11, 2026 · Curated by George's Blog
AI can generate 20 good Amazon images before lunch.
But which one would a shopper actually choose?
That’s where humans still matter.
When people hear “AI needs humans,” they usually picture someone writing prompts, fixing designs, and approving every version.
I think there’s a much more scalable role for people:
Let shoppers tell the system what works.
That’s why we’ve integrated PickFu into Listing Optimization AI
You can put your creative concepts in front of real people and get their feedback.
Which image catches their attention?
What makes the product easier to understand?
What feels confusing or hard to trust?
Their reasons can be more useful than the votes.
“I couldn’t tell how big it was.”
“The other image made the benefit clearer.”
“This one looks fake.”
Now you have direction for the next version.
Generate. Get feedback. Improve. Test again.
The part that excites me is where this can go:
A system that runs the process autonomously, with real human feedback guiding each round.
You don’t need to grow a creative team at the same pace as your product catalog. You need a repeatable way to learn from the people you’re trying to sell to.
AI scales the production.
PickFu brings the human perspective.
Our vision for ListingOptimization.ai is to connect those two into a continuous improvement loop.
Humans are still part of the process.
Their role is to help it learn.