The AI Outlook With Craig Planson of Browzwear

Released in The Interline’s AI Report 2025, this executive interview with Browzwear is one of an eight-part series that sees The Interline quiz executives from companies who have either introduced new AI solutions or added meaningful new AI capabilities into their existing platforms.

For more on artificial intelligence in fashion, download the full AI Report 2025 completely free of charge and ungated.

Where do you believe we currently are on the progression curve from AI as an extremely broad set of capabilities and promises, to AI as the foundation for applications and services that can deliver a measurable return on investment in well-defined areas?

AI is currently best positioned to drive productivity where there are repetitive, mundane tasks. Today, there is still a lot of confusion about where AI has the most significant impact. Some simpler use cases with AIGC that are getting the fanfare are becoming quickly commoditized. The real AI impact will be found by truly increasing productivity and efficiency at the operational layer. The progression will be realized in phases, and organizations must have a plan.

Since so many AI strategies have focused on the use of generative models to inspire and aid in design, we would expect these use cases to perhaps be the best-developed and the most mature. But right now it feels as though, whether it’s for technical or cultural reasons, there’s less alignment on the right approach, and fewer examples of best practices, than we might expect. What’s your take on where AI should really be in the design process? Is there anything you believe the industry is overlooking, or framing incorrectly?

The design ideation layer is clearly the first mover use case. The challenge is that the rapid development of ideas is creating a bottleneck for the product development phase. Change is driven by leading people through the process, clearly showing them what is in it, and having the right structure for success. The fastest starting point is to change the policy, build the process and tech around it, and address cultural resistance. We have found that having a north star is critical. A policy declaration around 95% first time right fit can shape the adoption and address the cultural resistance. Think about the time gained by avoiding 2 sample runs per style. This is more than chasing ‘AI first’; it is about having a plan and trusted advisors to guide you on the path.

Technology in general often treats “design” as a single unified discipline, rather than as two complementary workstreams: one focused on ideation and creation, and the other on more technical workflows. That also seems to be a risk where AI is concerned, with generative tools for inspiration becoming conflated with models and tools for the purposes of precision engineering. How important do you think it is to keep those use cases separate from both a technical and a cultural point of view?

We really need to unify the efforts while maintaining the domain expertise. When everyone is working towards the same goal and understands the overall issues of the organization, real ROI is achieved, and the organization can separate itself from the pack. There is work to do in this area to bring down the silo’s while maintaining integrity with the process. The industry has a bit of a practice of throwing their problems to another group or their vendors. This is why we need a North Star and KPI’s that drive integrated outcomes.

Even before the current AI boom, it’s been a challenge in 3D strategies to maintain the association between high quality visualization and pattern-accurate simulation. As a result, we’ve all seen roadblocks emerge where some digital product creation pipelines and workflows have produced fantastic-looking content, but with that fidelity coming at the expense of producibility. Given some of the concerns that have plagued generative AI from an accuracy perspective over the last couple of years, this seems like it could become a more significant problem where products start as AI generations, without the right grounding in reality. What do you see as the solution to this? What is it going to look like for brands and their partners to take advantage of the capabilities of AI, but in a way that’s anchored into the realities of product development and production, so that what they end up with is a representation of a product they can trust is really manufacturable? 

Excellent question. It will be critical that the AI-generated visualization is turned into accurate patterns and designs on which decisions can be made. The product development and accuracy requirements can be addressed with assistance from AI. We believe that a trainable AI asset is critical. The data and reference images inform the automation. Garbage in and garbage out with assets that lack accuracy and integrity. When you train the models on assets with real measurements and the digital informs the physical, you can truly exploit AI. Ultimately, we must be able to reduce friction in converting ideation into producibility. 

When concerns about the utility or the legality of using big, general-purpose AI models, trained on a gigantic set of data that spans different industries, brands, and sources, come up, the answer is always “use your data instead”. Realistically, what does that actually look like? How can a brand approach the task of deploying AI for design and development purposes that’s properly grounded in their own knowledge, heritage, products, and so on? And how can they use that deployment approach to differentiate themselves from competitors who might have gone down the off-the-shelf model route?

It is essential that you do not train AI models on the entire internet. There are many reasons, from unknown legal and ethical perspectives, to create a segregated instance that is not entangled with public reference images. While the segregated approach limits some AI training breadth, organizations can ensure they comply with brand philosophy and mitigate other risks, including protecting proprietary assets. When it comes to patterns and accurate virtual twins, the data and reference images need to be developed and certified for fit. This is a small language model approach that begins to enable automation of the accurate virtual twins, where real gains are realized. I’ll go back to the north star of 95% first time right fit. When execution is simplified, the whole process improves.

What do you believe are the next steps for how AI is deployed and used? Is it more likely that AI will solidify its place as a new human interface paradigm, the frontend of tools and workflows? Or is its future closer to what cloud infrastructure has become today – a quieter commodity that is still the foundation for the next generation of applications, but in a less obvious way than what we’ve seen over the last couple of years? Or is it both?

I would say it is a combination of both. Some use cases, like ideation, are clearly commoditized, and that will accelerate. We believe that accuracy and an open approach to platform integration are critical. Organizations need an environment where they can integrate with best-of-breed solutions without limitations. The more compelling use cases are realized in maintaining proprietary product assets in a secure environment and unifying the process. Overall, people who use AI to be more productive will be highly relevant and valuable assets in the marketplace. Use cases are accelerating, but key distinctions need to be understood and planned for.