AI Just Figured Out How to Stop You From Returning Clothes. It Took $200 Billion in Losses to Get Here.
Virtual try-on AI finally hit ROI positive. The fashion industry just found the first AI use case that pays for itself on day one.
The AI Post newsroom — delivering AI news at the speed of intelligence.
The fashion industry has a $200 billion problem that nobody outside retail talks about: returns. Online shoppers buy three sizes, keep one, send two back. The logistics cost alone is eating retailers alive. Margins that were already thin are turning negative.
Now AI is finally solving it. And the math actually works.
CNBC reports that AI startups like Catches are deploying virtual try-on technology that lets shoppers see exactly how clothes will look on their body before buying. The key breakthrough is cost. Running AI-powered visual rendering used to be prohibitively expensive. The advances in inference efficiency over the past year brought the cost down far enough that the ROI is now positive from day one.
"The returns problem is solvable now due to advancements in AI, allowing firms to run visuals for end users cheaply enough to make a return on investment," Catches CEO Ed Voyce told CNBC.
This is quietly one of the most important AI stories happening right now. Not because virtual try-on is flashy. Because it is the clearest example of AI crossing the line from "cool demo" to "immediately profitable deployment." No hype cycle. No five-year roadmap. Just a real business problem meeting a real AI solution at a price point that works.
The timing matters. While the tech industry argues about whether AI will achieve superintelligence or replace all human workers, fashion retailers are quietly deploying AI that does one specific thing really well: show you what a dress looks like before you buy it. No existential crisis required.
For context, retail returns cost the US economy roughly $200 billion annually. Return shipping, restocking, quality inspection, lost inventory. If virtual try-on cuts return rates by even 20%, that is $40 billion in value unlocked. At scale, this becomes one of the largest measurable economic impacts of consumer AI.
The lesson for the broader AI industry: stop chasing AGI headlines and start solving expensive problems. The money is not in building god. It is in stopping people from returning jeans.