AI is collapsing the time between inspiration and execution. As creative tools, digital product development and consumer intelligence accelerate from both ends of fashion, the race is increasingly in the middle, where digital ideas become physical products.

Gone are the days of painstakingly creating flat sketches and colorways one by one for a line sheet. I tested FLORA ’s Fashion Studio by uploading an old dress design and giving it a single prompt. Within seconds, it generated a line sheet with five colorways, front, back, and side views, product descriptions, design details, and suggested pricing.

I could alter the model’s hair color, change colors, and regenerate views without rebuilding the work from scratch. Tasks that once required a designer or merchandiser to manually create and revise flats, colorways, and presentation materials could now be accomplished with an image and natural-language instructions.

That’s when the larger question became harder to ignore.

If fashion can now create and iterate at this speed, what happens when we decide to actually make the dress?

FLORA can generate five colorways in seconds. A mill can’t dye five production runs in seconds. It can visualize the back of a dress instantly; a pattern maker still has to determine how that back is constructed. And once a brand commits to materials, colors, quantities, and factory capacity, changing its mind becomes considerably more expensive.

On screen, fashion is moving at AI speed. The physical supply chain isn’t.

Building A Creative Environment For The AI Era

Wong began building FLORA while in art school, frustrated that the first wave of generative AI tools could produce impressive outputs but lacked the control and workflow professional creatives needed.

FLORA’s Canvas approaches the problem differently. Its node-based workspace allows text, images, and video to be connected into creative workflows, with one output feeding another and ideas branching without losing the process that produced them. Rather than requiring creatives to constantly switch between individual AI products, FLORA brings dozens of generative models into a single environment.

The distinction is increasingly not simply which model produces the best image, but how those models are organized around the way creative people actually work.

Fashion emerged organically as one of FLORA’s strongest use cases. Wong said that after the platform launched, it began attracting teams across creative industries, from Pentagram and IDEO to Nike, with particularly strong adoption in fashion. He attributes some of that early traction to generative models becoming especially capable at fashion imagery, allowing designers to explore ideas and variations quickly.

Power users within companies including Prada and Jordan Brand were already building sophisticated workflows in FLORA’s Canvas. But bringing that same capability to broader fashion teams presented another challenge: not everyone wanted to navigate nodes, select models, or construct complex workflows.

That need ultimately led to Fashion Studio, a simpler interface designed to handle key tasks throughout the design and production process—from initial sketches and renderings to fabric choices, model try-ons, and full campaign photoshoots. “Each step takes what you already have and gives back what you asked for,” Wong wrote in a LinkedIn post announcing the feature, describing an end-to-end workflow that transforms concept sketches directly into finished campaign assets.

From One Idea To One Hundred

The more consequential change may not be how quickly AI produces one finished image. It is the number of possibilities a creative team can explore before deciding which one deserves to move forward.

The value of AI isn’t simply speed to completion. It is speed across possibilities.

But as execution becomes easier, Wong argues that something distinctly human becomes more valuable.

“If execution gets faster, your taste actually becomes more and more valuable,” he told me.

Generating 100 possibilities does not determine which silhouette belongs in a collection, which color is wrong for the customer, or which image actually expresses the brand.

And that raises another question: What happens downstream when a team that once developed ten possibilities can suddenly explore 100?

When An Image Has To Become A Garment

I asked Wong whether accelerating the front end of fashion actually shortens the overall product calendar or simply pushes a larger volume of decisions downstream into product development, sourcing, and manufacturing.

He sees the tech pack as one possible bridge between those worlds.

“What you make in the tool can immediately become a tech pack,” Wong said. “That’s very solvable.”

For FLORA, that remains an opportunity rather than the core of its current Fashion Studio proposition. Other fashion technology companies are already pushing further into this digital middle.

Style3D has built a broader ecosystem connecting AI-assisted ideation with 3D garment development and technical product information through digital assets that can ultimately be reused for marketing.

Its CEO, Eric Liu, has an evocative description for what could happen as AI connects fashion’s traditionally sequential digital processes: a “wormhole.”

In an interview with Just Style, Liu described a future in which AI could collapse the traditional progression from planning and design to production and sales, bringing fashion closer to an “instant fashion” model in which an idea could move toward sales-ready materials in hours or days.

But Liu draws an important boundary at the point of physical production.

“Of course, for production, it’s another story. It’s still very difficult,” Liu told Just Style, adding that logistics and production remain “quite a long way to go” in adopting AI.

Read Eric Liu’s interview with Just Style

Even as technology connects more of the digital workflow, there remains a critical translation between what looks right on a screen and what can actually be manufactured.

Richard Zhao, Supply Chain Director at Peerless Clothing Inc. and former President of Smart Apparel (US) Inc., sees that divide from the manufacturing side.

“AI makes beautiful images. It does not make a pattern,” Zhao told me.

Zhao points to a potential gap as younger designers increasingly work with generative AI. Designers without deep technical knowledge of textiles may not always recognize the difference between a compelling visualization and the information a mill actually needs to manufacture it.

A textile CAD or technically accurate fabric reference doesn’t simply show what a fabric should look like. It communicates the underlying construction, the yarn, structure, colors, and other details a mill needs to develop a handloom. An AI-generated textile design, by contrast, can look remarkably convincing on screen without containing that underlying construction information.

Zhao described how the development had to be corrected using an actual swatch or technically accurate reference the factory could follow, adding back time and resources that the digital process was supposed to save.

That is the unfinished work in fashion’s AI middle. The challenge isn’t only generating something that looks manufacturable. It is translating creative intent into technical information that can be manufactured the first time correctly.

More possibilities, therefore, don’t automatically create a faster fashion calendar.

As Zhao put it: “More options are not the same as better options. Somebody still has to say no, and saying no takes judgment and time.”

Because the middle is where digital possibility becomes physical commitment: where a color becomes dyed fabric, a rendering becomes a pattern, a forecast becomes a purchase order, and an idea begins consuming real materials, factory capacity, and capital.

The race is to connect them.

Moving Decisions Out Of The Critical Path

Zhao says AI is already eliminating meaningful time inside supply chains, but much of that work is happening in the office rather than on the sewing floor.

Factories are using AI to analyze and re-sequence order demand, rebuild production schedules, and run costing assessments, he said, reducing the back-and-forth that once took days.

Automation is advancing on the factory floor as well, including robotic movement between operations and programmed sewing units for repeatable tasks. But Zhao distinguishes much of that from AI.

“It helps efficiency. It doesn’t move the critical path much,” he said.

In technically complex products such as tailored clothing, skilled operators still rely on something machines have struggled to replicate: the human touch.

“You can’t program touch yet,” Zhao said.

So how does fashion get from a 90-day calendar to 60 days, or even 30?

Zhao argues that the answer may not be to make physical production dramatically faster.

It is changing when the decisions are made.

He points to a model he worked with firsthand while serving as President of Smart Apparel (US) Inc. Rather than committing all finished inventory upfront, Uniqlo would develop and confirm the style and colors, place an initial launch order, and reserve yarn for subsequent production.

Then the market could provide more information.

Store and marketing sales data informed demand analysis by color and size, enabling subsequent purchase orders to reflect what customers were actually buying.

Because the yarn was already positioned, the colors were approved, and the samples were completed, the remaining production cycle could move much faster.

“The mill produces fabric in about 10 days; yarn is in hand, colors are already approved, so it’s dye and weave,” Zhao explained. “The factory needs another 10 days because every sample is already approved. Ten plus ten.”

Instead of taking the risk in thousands of finished garments in potentially unwanted colors or sizes, more of the risk remained upstream in yarn, where it was cheaper and more flexible.

That changes the question from “ How can AI make a sewing line faster?” to “ How can AI help fashion make better commitments sooner?”

This is also where consumer-facing AI becomes relevant to the supply-chain conversation.

In my recent reporting on Zelig’s work with Revolve and Lulus , tools such as Build a Look and Shoppable Closet showed how AI can create new interactions before and around the transaction: what shoppers combine, consider, save, and ultimately purchase.

Those signals don’t manufacture a garment. But as AI becomes better at interpreting consumer intent, it could provide another source of information for upstream decision-making.

And that suggests a different way of looking at the AI transformation of fashion.

AI is attacking uncertainty from both ends of the industry.

On one end: What could we create?

On the other hand, what does the customer appear to want?

Between them sits the much harder question:

What should we actually make, how much should we make, and when should we commit?

When Execution Gets Easier, Judgment Matters More

There is an interesting connection between Wong’s creative world and Zhao’s manufacturing one: both ultimately arrive at human judgment.

When I asked Wong whether AI could diminish the role of designers and creative directors, he drew a distinction between creativity and execution. In his view, AI isn’t eliminating creativity; it is reducing the human labor required to execute creative ideas. As execution becomes faster and more accessible, the value shifts toward the ideas themselves—and the taste and judgment to know which ones are worth pursuing.

His advice wasn’t simply to become better at prompting. He emphasized visual language, composition, art and design history, cultural references, and the ability to recognize strong work.

Knowing what to ask for—and knowing whether the result is any good—becomes more important when generating another alternative costs almost nothing.

Zhao adds another dimension: technical judgment.

Creative judgment determines which idea deserves to move forward. Technical judgment determines whether and how it can actually be made.

And there is one final constraint that technology alone cannot solve.

Good factories need predictable orders. Greater flexibility for brands has consequences if suppliers aren’t given better visibility into what’s coming.

“If brands use AI to get more flexible without giving factories more visibility, the speed just gets paid for by somebody else,” Zhao said. “That’s not a technology problem.”

Fashion has always been both an idea and an object.

AI is making the idea extraordinarily fast. It is beginning to connect previously separate digital processes and provide better information about what consumers may actually want.

Now the race is to the middle - to connect creative intelligence, technical development, consumer demand, and physical capacity early enough to make better decisions before expensive commitments are made.