On paper, the digital thread is simple: every system talks to every other, and product data moves from design through to production without getting lost along the way. In Asia, where most of the industry's product actually gets made, you find out fast which parts of that picture hold up in practice.
That was the brief for our latest PI Spotlight. Bernd Sauer (B_CONNECT, ex-Puma & adidas) chaired, joined by Joyce Wong (Top Form Group), John Dulip Kumar (PVH) and Alex Thomas (Caleres). Combined, that's well over a century of manufacturing, sourcing and product creation experience on one panel.
Over the course of the hour they got into how close AI is to taking a product from a phone photo to a factory-ready tech pack, why some suppliers still order double the material "just in case", and why a brand should be glad when its manufacturer knows more than it does.
🎥 Watch the full discussion below & read on for the full write-up.
What Does "Success" in the Handoff Actually Look Like?
For John, success is when the factory gets the digital intent once and builds it correctly the first time. In business terms that means fewer prototype and fit rounds, lower development cost, quicker approvals and fewer nasty surprises in bulk, which all adds up to getting to market faster. If you want something to measure, he'd look at sample-round reduction, first-pass fit approval, tech pack completeness, bulk defect rates and development cycle time.
He sees the promise falling short in three places. Data quality comes first: plenty of companies run digital processes on top of incomplete PLM records and patchy tech packs, so factories still have to come back with questions. Then interpretation, because two factories can get identical files and still ship different products. And finally, accountability. When something goes wrong, the problem usually turns out to live in the gap between design, sourcing, quality and manufacturing, where nobody quite owns it.
The industry has largely digitised the handoff. What we haven't fully mastered is industrialising the execution. The digital thread succeeds when a factory can build the product exactly as intended without asking the question twice.
— John Dulip Kumar, PVH Far East Limited
Joyce sees it from the other side. Some of Top Form's customers still send a sketch, ask the factory to build the 3D, and then ask for a physical sample on top. That's the same work done twice. Once the file goes up onto the customer's platform and gets modified, the factory never hears what happened to it. Her version of success is simpler: the brand sends the 3D file, sample rounds come down, and the first sample is the right one.
Alex's concern was material physics, which he said is a bigger struggle in apparel than in footwear.
Success isn't a perfect rendering with perfect lighting. It's getting the physical attributes right, tensile, shear, mass, directly into the geometry file, so it becomes a digital twin faster. That background work isn't always visible, but it makes a big impact on the final output.
— Alex Thomas, Caleres Inc.
Supply chain lives and dies by KPIs, he added, and some companies now track a digital sample acceptance rate (DSAR), aiming for something like 85% of samples approved digitally before anyone cuts a physical one.
Where 3D Earns Its Place & Where It Still Falls Short
Top Form makes close-to-body products, where a lot of the value is hidden: construction and technology that a flat sketch or a CAD image can't really show. Joyce gave three places where 3D pays its way. Seasonal concept presentations are one, since silhouettes, colourways and that hidden construction all sell better in 3D, and adoption rates go up. Costing is another. Some customers ask for 3D early so they can compare options side by side, each with a price, and decide before a single physical sample exists. But the most common use is also the plainest: checking colour and material before cutting, which catches mistakes before they cost a sample round.
Where it falls short is the gap she'd already flagged. The factory builds the 3D at the customer's request, often on top of a physical sample, and then has no idea what the file is used for next. Without knowing, it can't tell whether the extra work fed into a decision or went nowhere.
Bernd was quick to add that 3D makes its own demands.
3D isn't something automated that comes out of the computer like a miracle. You still need a skill set, and there are quality standards you have to follow.
— Bernd Sauer, B_CONNECT
That came through in an audience question on how accurately 3D can grade a style across a full size range. Alex said sample-size avatars are fine, but grading automatically out to the extreme sizes still takes a lot of manual work. In footwear, his world at Caleres, every size means deciding which parts of the shoe stay fixed and which are allowed to scale, and those calls have a big effect on cost and scalability. That's judgement the software can't make on its own yet.
The bigger variable, John said, is mindset. Plenty of people still only trust physical samples, because apparel comes down to touch, feel and fit. Going all-in on 3D isn't the answer either. He wants brands, designers and product developers to make it part of how concepts get worked up, a tool rather than an either/or. In his view it falls through the cracks when people expect too much of it, or too little.
Is "AI Versus 3D" the Wrong Frame?
Bernd's view was that one doesn't exclude the other. AI is quick: type a prompt and you have ideation images, storyboards or sales imagery in minutes. But what it still can't give you is product accuracy. Validated engineering data, exact logo placement and grading that holds up all still come from 3D, built from real patterns and stitched together properly.
His advice was to get that foundation in first. Most brands have already digitised their basic blocks in 3D, so start there, then use AI to try things out and test ideas on customers before anything goes near production.
Alex agreed it isn't a contest today, but then told a story that suggested it might not stay that way for long. One of his tier-two partners runs a lighting factory in China and has hooked Claude up to his bill of materials through APIs. He takes a photo of a Perrier bottle on his phone and asks for a lamp based on it. Claude produces the image, then a bill of materials drawn from his real inventory (six-millimetre glass, a two-millimetre brass holder, a three-millimetre canopy), then a step-by-step construction chart and finally a tech pack. He walks that onto the factory floor.
He has an end-to-end workflow just by connecting Claude to his bill of materials. I do believe it's only a question of time before one eclipses the other.
— Alex Thomas, Caleres Inc.
Soft goods are harder, he conceded, and getting fabric to behave will take longer. He handed that point to Joyce, whose team at Top Form has far more experience with fabric. Their workflow today is deliberately hybrid.
It's quite difficult to use prompts to describe or generate an image of our product very precisely. So the first step is to build the 3D based on our pattern and all the construction details, then plug that rendering into AI to generate a storyboard.
— Joyce Wong, Top Form Group
Customers find that far more engaging than a CAD image on a body, Joyce said. With activewear they go a step further, using AI to animate the garment so customers can see it working in motion instead of imagining it from a still. That, she said, is what gets the "wow".
John agreed with them both.
They're not competing technologies, they're complementary capabilities. 3D provides the digital representation of the product; AI helps us make better decisions with that representation. So it's not AI versus 3D. It's AI-enabled 3D.
— John Dulip Kumar, PVH Far East Limited
What's at Stake When Supplier Data Isn't Standardised?
Alex named two costs. The first is margin of error. The handoff isn't yet clean enough for a factory to take a digital marker, yield and pattern at face value, so during early prototyping some suppliers buy double the material, just in case. The second is compliance. Digital product passports are coming, and if supplier data is missing or wrong, the worst case is a shipment held at the port of entry.
John went further, because at PVH's scale, data is the business.
The currency of commerce is not dollars anymore. The currency of commerce is data.
— John Dulip Kumar, PVH Far East Limited
When data isn't standardised across tier-one, tier-two and sub-tier suppliers, the same factory or material source can turn up under different names in different systems. Then you can't see your real exposure, capacity or dependencies. When something goes wrong (a geopolitical event, a raw material shortage, a quality problem) leadership needs accurate numbers straight away, and instead teams end up cleaning and reconciling data rather than acting on it. Collecting it isn't the hard part, John said. Trusting it enough to act, and to put your name to a sustainability or compliance claim, is.
Bernd's caution was that piling on more systems won't fix it. Sometimes less is more.
When the Manufacturer Knows More Than the Brand
For John, a manufacturer knowing more than the brand is simply how things are now, and a good thing. The days of brands assuming they had all the answers and treating suppliers as converters are over.
The best Asian manufacturers bring real depth in materials, manufacturing engineering, automation, sustainability and digital product development, and they often spot a production problem (and the opportunity in it) before the brand does. So the brands doing well are moving from buying to co-creating. That means sharing information earlier, bringing suppliers into product decisions sooner and rewarding them for innovation, not only for compliance.
The competitive advantage of the future may not be about who owns the best brands, but who builds the strongest innovation ecosystem with their suppliers. The best manufacturers are no longer vendors. They are extensions of the brand's product and supply chain teams.
— John Dulip Kumar, PVH Far East Limited
Bernd has lived in Hong Kong for 19 years and called the change in supplier capability over that time mind-blowing. It would be foolish, he said, not to use it. Joyce agreed that it starts with partnership rather than a list of demands. Factories understand garment construction and how materials behave, so they know what can actually be made.
At the factory side we can propose many different options that are workable for the factory and also sellable for the brand. That adds value before the decision is locked down.
— Joyce Wong, Top Form Group
When Traceability Data Is Wrong, Who's Accountable?
The brand carries the reputational risk, and the tier-one vendor has the financial liability. Even if a failure occurs at tier three, like a yarn spinner, ultimately the financial relationship is between the brand and the tier-one vendor.
— Alex Thomas, Caleres Inc.
Beyond the contract, it's up to the brand how far it goes on the money side in the name of partnership, and whether a failure becomes a claim. The underlying problem, as Alex sees it, is a data vacuum. Brands tend to want all the data without paying for it, because they treat it as their secret sauce.
The rules keep moving too. Europe is out in front on compliance and traceability, with the US not far behind. Partners in China are putting serious effort in as well, and for some products the requirements for the Chinese domestic market now go further than those for the US. There's no short answer, Alex admitted, and it isn't getting any easier.
Breaking Down the Silos
An audience member asked how to break down silos between stakeholders. For Alex, it comes down to incentives.
What are the incentives to collaborate? If you're being rewarded for being in silos, you'll do more of it. If you're incentivised and rewarded for breaking them down, that's what you'll do more of.
— Alex Thomas, Caleres Inc.
He'd measure it, too: how many engagement points a team has, how many ideas it's picked up from outside its own area. He pointed to the circular building at Apple's headquarters, designed so people run into each other. That kind of chance encounter has to be designed in, he said, and it has to come from the top.
Bernd agreed about leadership. You need a clear "lighthouse" vision, repeated often, and regular cross-functional check-ins. John called silos an age-old problem that only shifts when someone deliberately steps in, and his tool of choice is shared KPIs. If you're only measured on your own function, you hit your number, throw the work over the fence and call it done. A shared KPI (did the right product reach the consumer, on time and at the right quality?) gives everyone skin in the game. His example was a development lead who stays accountable for what happens in production, even though the production team is the one doing the work. You control what you can and influence what you can't.
Build or Buy?
Hybrid again. Alex's rule of thumb is to buy where experts have already done the legwork, such as CAD, 3D simulation and PLM, because there's no point reinventing the wheel. And to build the pieces that join things up: custom APIs, middleware and your own data pipelines linking a digital showroom, an ERP or a costing engine.
The digital asset management system often ends up on the build side, because the way a company catalogues, reuses and pushes assets out to B2B and B2C channels tends to be specific to that company. Customisation has a price, though. Every upgrade can eat weeks, and retooling can run to millions. Lightweight cloud tools, he added, help tier-two and tier-three suppliers scale faster, since there's no legacy system in the way.
Bernd cared less about where a platform comes from than whether it talks to what you already have. One that plugs into your existing systems is an advantage; one that only works with its own sub-platforms is a risk. Joyce described that mix at Top Form, which built its own core system but also works inside whichever platform each brand customer uses.
Companies should buy the foundation and build the differentiation. The real question isn't whether to build or buy. It's deciding what creates competitive advantage and what doesn't.
— John Dulip Kumar, PVH Far East Limited
Build everything yourself, he said, and the technology has moved on by the time you're done. Buy everything and adapt to it, and you've only standardised what already exists.
If the Technology Exists, Why Are People Still the Barrier?
Joyce's answer applies to brands, agencies and factories alike. People have different roles and different goals, and hardly anyone is measured on adopting 3D. If the old way works, why change? A lot of people also don't know much about 3D, which makes them nervous of it, and learning it takes time.
Top Form's fix is a cross-functional 3D task force with leadership behind it.
We built our 3D task force with members from every function team. At a monthly meeting we share case studies about using 3D for different customers, and digital news from the market, so every function knows what's happening. That helps them understand more about 3D and be less afraid of it.
— Joyce Wong, Top Form Group
After a year or two, teams started using 3D, nudging colleagues to use it and seeing results.
John thinks part of the problem is perception. Technology gets seen as replacing people's judgement, when the point is to take repetitive work off their desks so they can make better calls. People don't adopt something because it's new. They adopt it because it fixes a problem they care about, so he'd sell the outcome rather than the tool. Instead of "we're implementing AI", say "we're cutting sample iterations by 30%" or "we're getting rid of manual tech pack validation". Resistance is a given, he said, and visible results are the quickest way through it.
Alex had a warning from experience. During transitions, some companies make teams do everything twice, in 3D and physically, and he's seen that end in burnout. It usually comes down to a lack of trust, which is why the interim steps need planning too.
Bernd remembers sending bills of materials to Asia by fax. The tools have come a long way since, he said, and the people have to keep up with them.
Bernd's own takeaways were practical. Regulation is coming, technology is moving quickly, and the answers are rarely black and white, whether it's 3D or AI, building or buying. Go in with an open mind, he said, and look for the version that works for everyone involved, from brand to tier-one to sub-tier.
There's reason for optimism in that. At Top Form, a monthly task force turned 3D from something people were wary of into something teams now encourage each other to use. A lighting factory in China is getting from a photo to a tech pack by connecting AI to its bill of materials. And brands like PVH are starting to treat their best suppliers as co-creators rather than converters.
The thread may not be finished, but the panel had plenty of proof that it can hold.
This spotlight was a sneak preview of the conversation we'll be continuing in person at The Fashion Tech Exchange Hong Kong on 4th November, where all of the panellists will be joining us. Built for senior brand, manufacturer and technology leaders working at the intersection of design and production, it's a day of honest, practical, peer-to-peer exchange.
