Interview with Kyle Stephens, CEO at KolAi Denim
Kyle worked on Innovation for Lee and Wrangler Jeans from 2016-2020. He received laser training from both Jeanologia in Spain and VAV Technologies in Turkey and regularly traveled to factories around the world to oversee innovative developments with laser finishing.
KTJ: First of all please let us know about KolAi Denim’s history and its success story in the textile sector.
KS: KolAi Denim started as an idea.
While working for VF Corporation (Lee / Wrangler) I became familiar with laser finishing of denim and was at the same time keeping up with the academic literature in machine learning and artificial intelligence. I realized that new machine learning models could be used to help automate laser design for denim. I presented this idea to a colleague and denim laser expert Aydin (at that time he was doing laser and wash development for 7 For All Mankind which was a VF brand). He thought the idea was fantastic and never forgot about it.deSeveral years later, the COVID pandemic broke out and I was no longer working for Lee/ Wrangler (who’s parent company had become Kontoor). Aydin reached out to me. “Remember that idea about automating laser design?” He said. “Let’s do it.” And that’s how KolAi was born.
I brought in another former colleague from VF, Jack Payne, to help with the technical / programming side of things, Aydin provided the data, and together we built software that automates laser design for denim. We incorporated in 2021 and called our company KolAi from “Kolay” the Turkish word for easy + A.I. for artificial intelligence.
We now have a growing team with customers all over the world and we continue to improve our product with feedback from our customers.
Also, our first software product is just the beginning. We have a lot of other ideas and other products in the works, all with the aim of making denim design and development faster, easier, and more sustainable.
KTJ: Please tell us more about your range of products and innovations.
KS: Our core product is a web-based tool that helps automate the design of laser files for denim finishing.
Without our tool, laser designers typically photograph a target garment and then edit and illustrate that photograph by hand with a photo editing program (like Photoshop) to create a file the laser machine can understand. This design process is time-consuming– requiring hours of time for one design – and requires a laser designer with much skill and experience for adequate results.
With our tool, a laser designer can simply feed an image of their target garment to our A.I. system and it will automatically generate a design that is 90% complete. The A.I. does the tedious work so the laser designer can focus on the fine details and perfect the designs. With our tool, laser designers can get excellent result in minutes instead of hours.
In addition to the A.I. tool, our online platform also has a suite of professional editing tools (like Photoshop) and some other denim-specific automation to make the laser designer’s life easier. The outputs from our system work with any brand of laser machine.
We have other software products in the works, but no others have been released yet.
KTJ: KolAi denim is one of the leader software in denim designing, What are the unique features of KolAi denim software that was chosen by denim designers and denim manufacturers around the world?
KS: The unique features are speed, ease of use, and flexibility. Speed and ease of use because the A.I. does most of the tedious work of making a laser design for you.
Flexibility because the system is entirely online – no need to download any software or plug in any special keys, you only need an internet connection. The system is also flexible because the outputs are compatible with any brand of laser machine.
We are also unique because we are the only company that has trained a machine learning model (technically, it’s known as a “deep neural network”) to learn how to draw like a laser designer. We trained the model with 10,000+ examples of good laser design!
KTJ: Where is your main export destinations and countries?
KS: We have customers all over the world – Turkey, China, Vietnam, Mexico, Colombia, Bangladesh, Pakistan, and the USA – and we are looking to expand. Any factory that makes denim dry process with laser could benefit from our software.
KTJ: As a textile software Expert, how do you evaluate the textile industry in the Middle East and Africa and its potential for the Future? How do you evaluate the kolAi denim market share in the region?
KS: Currently our biggest customer base is in Turkey and Turkey is well-known for their expertise in laser finishing denim. We have several technicians working for us in Turkey and they service much of our customer base. I think Turkey will continue to play this role in the textile industry – exporting knowledge and expertise to factories in other parts of the world.
We have only recently begun conversations with manufacturers in North Africa and the Middle East (Jordan in particular). From what I can tell based on these conversations, these regions are growing and should continue to grow based on the quality I’ve seen and their proximity to the EU.
KTJ: How Artificial Intelligence can help the textile industry in general?
KS: This is a great question and I just gave a talk about this at the Denims and Jeans Expo in Vietnam.
I think the sky is the limit with how A.I. can help, especially with these new ultra-powerful A.I. models that have been coming out recently (things you have probably heard about in the news like ChatGPT and Midjourney). We have already built one tool to automate denim laser design. Here are a few more examples of A.I.-based tools that will likely come out soon:
• A.I.-Assisted Digital Design. Soon, denim designers will be able to input the kind of information they might include on a mood board (inspirational imagery, sketches, fabric swatches, trims, etc..) and an A.I. will be able to consolidate this information and generate realistic imagery of what that the resulting garment would look like. Designers will then be able to interact with this imagery using prompts (e.g., circling the pocket and instructing the A.I. to change its style). This will enable a lot more digital design and less sample-making.
• A.I.-Assisted Physical Garment Creation. Of course, the hard part is turning a digital design into a physical garment. A.I. can help with this too. For example, determining wash recipes or dry process laser files for digital garments.
These tools are specifically for denim design because that is the world that I am familiar with. However, I’m sure that A.I. tools could also help with other aspects of the textile industry (supply chain, etc.), I’m just not as familiar with the problems in those areas. If you have an idea or a problem that you think A.I. could help you solve, please reach out! We are very much open to collaboration.
KTJ: How do you evaluate the textile and apparel sector in the USA?
KS: We have several customers in the US who find great value with our platform because of their high labor costs. The US is still the center for many international brands and they will continue to run design and some development/sample-making out of the US. The key is to pass their developments and know-how to overseas manufacturers effectively. One of our goals to help make this transition as smooth as possible with software and automation tools, to reduce the amount of sample-making and around-the-world shipments.
KTJ: You can mention if you have any local or international standards and certificates.
KS: There are not any standard certificates for software that I know of, but we are helping
with sustainability by reducing the number of sample garments that need to be made
and we will soon be implementing our own certification program for laser design to
ensure a certain level of quality among laser designers.
KTJ: At the end, you can add any other information
KS: We will be at ITMA in Milan in June, so come say hello if you would like to chat in person!
 I became familiar with laser finishing of denim and was at){kind=link}