Type “cycling jersey, geometric alpine pattern, teal and copper” into any image model and you will get something gorgeous in nine seconds. There are a dozen AI jersey generators online. They are free. They are fun. They are also, from a manufacturing standpoint, completely useless.

Because a jersey is not an image. A jersey is a set of panels that get cut, sewn, and worn — and the pattern has to survive that journey.

The graphic has to cross a seam and still line up on the other side. It has to grade from XS to 3XL without the sponsor logo stretching into a smear. It has to sit at the right resolution when it hits the sublimation printer, and the colours have to come out of the heat press looking like what you approved on the screen, not eight percent duller. The gradient that looked lush in the render has to not band. The fine linework has to not disappear into the fabric.

None of that is a design problem. All of it is a manufacturing problem. And AI image tools do not solve manufacturing problems, because they have never met a factory.

So the industry did what industries do. It kept the AI at the front, where it demos well, and quietly put humans at the back, where the work actually happens. Upload your prompt. A designer will review it. We’ll get back to you on whether it can be made.

That gap — between the prompt and the panel — is where every one of these offers currently lives. It is a human, at a desk, redrawing your AI dream into something a printer can eat.

We spent twenty-two years on the wrong side of that gap

I have been manufacturing custom sublimated sportswear since 2002. Long before any of this, I was on the receiving end of the “here’s my design” email. I know exactly what arrives, and I know exactly what it costs to fix.

That is not a boast. It is the reason FLAVE exists, and it is the reason I think most people building in this space are building the wrong end of it.

If you come at this from AI, you build a magnificent prompt box and then discover that fabric has opinions. If you come at it from manufacturing — if you have spent two decades watching designs fail on the print bed for reasons that are obvious in hindsight and invisible on a screen — you build the panel engine first, and the prompt box is the easy part you bolt on at the end.

FLAVE generates production-aware files. Not a picture of a jersey. The actual panels, with the seams understood, graded, and ready to go to a printer. That is the whole company. The prompt is the interface. The output is the product.

So we tested it the only way that counts

A claim like that is worthless in a demo. Any file looks production-ready until a production line disagrees with it.

So we sent FLAVE-generated jerseys — from prompt, end to end, no human redraw — to seven factories, on four continents:

Bangladesh. Brazil. Bulgaria. Poland. Vietnam. Thailand. Pakistan.

Different machines. Different operators. Different inks, different fabric lots, different calibration, different languages, different quality cultures. Seven independent chances for the file to be wrong.

Seven jerseys came back. All of them printed.

I want to be precise about what that proves, because precision is the entire point of this post. It does not prove FLAVE is finished.

FLAVE is in pilot. 900 brands have signed up to the waitlist. Our production network is real but still expanding — we are adding nodes, and the map is not yet the map I want it to be. If you sign up today you are early, and early means rough edges.

Sign up to be one of the first to get your hands on the industry’s 1st AI solution for sportswear brands that allow you to produce globally, fast.

Kevin Van Lancker founded Sportswear of Tomorrow in 2002 and is the founder of FLAVE Inc.

FLAVE is currently in pilot. Join the waitlist →