Last month, Duluth’s most ambitious startup rolled-out a new artificial intelligence product called Fashion Brain. This month, we spoke with cofounder & CEO Markus Müller regarding the development and company prospects:
When did Tryon Media start and for what purpose?
We began in 2010/2011 by my wife Lara Signorelli and myself to offer distinctive fashion publishing resources. She was able to start working on the company before me, so technically I was her first hire…it only makes sense since she has the degree in photojournalism and I am a college dropout (laughing).
We work with editors, brand managers, retailers, by offering them custom content – stories, videos and photos. Historically, our customers would work with us on their print publications, though as you can imagine this is rapidly becoming a digital universe. The scope of our offerings has evolved and we’ve jumped in artificial intelligence and are rolling out the technology on the red carpet with Fashion Brain at the Oscar Awards.
How does it work?
Publishers today are tight on resources, expected to do more with less, and experiencing faster deadlines. So all the metadata we have been compiling over the years around our content archive is now being leveraged through our Fashion Brain artificial intelligence to do what we’ve always been doing – but faster – for them.
The AI can determine the photo characteristics (full length, portrait, etc.), and goes into the subject matter such as their hair or clothing color, style such as v-neck, jewelry or dress length, etc.
For example, People Magazine received over 100,000 images all in one night from the Oscar’s event. So if they spend even 3-5 seconds per image, it will take staff a long time to parse all of them properly. By having our AI look through the images at first, it can spec them out and significantly expedite this process.
Then comes the real, almost magical capability — when Fashion Brain extracts certain styles which are a combinations of these things and suggests trends and stories. There are also nuances such as the ability for the AI to determine authentic vs. knockoff products, something that’s appealing to brands and manufacturers.
What is unique about your artificial intelligence aka Fashion Brain?
Our competitive advantage is years of curated content key worded by humans that we are using to train our models. This archived repository is so deep and accurate that it cannot be beaten by sheer quantity or brute force; 500,000 quality images with proper metadata will handily beat 500,000,000 images without.
How do you benchmark your own Fashion Brain AI vs. others, and how does it currently compare?
Right now it is focused on image recognition, and in the future we will be adding natural language processing.
We benchmark FashionBrain in two ways: Firstly, we’re interested in testing how it performs compared to general purpose platforms like Google Vision or Clarifai, as well as special purpose platforms for fashion tagging, for example VueTag or Tagwalk.
Even in its current very limited beta release FashionBrain beats all of these platform handily.
Secondly, we compare FashionBrain against our own team of highly trained and experienced fashion editors at Runway Manhattan. For several years Runway Manhattan has been the gold standard in fashion keywording — no other fashion imagery provider has any standards or results that compare even close. That’s the real standard we measure FashionBrain against and that we have to compare with. And we’re getting pretty close already.
Once we add in the NLP, then we will have the ingredients to delve into predictive analytics. In fashion, developments such as the time of year, colors, a region, or even materials sourcing can all be forward looking.
Are customers paying for Fashion Brain technology at this time?
Not yet – we are letting people test it to tell us what they think and welcome critique.
How many and who are some of your current paying customers under your traditional model?
Around 300 direct clients who have bought from us over the past 12-24 months. Time, Mondadori, Conde Nast, Reuters, Chanel – among many domestic and international brands, retailers, and publishing houses.
If Fashion Brain goes according to plan, what potential impact are you projecting?
There are three paths. The least exciting — but at the core of what we set out to do if nothing else — we can do what we are already doing better by augmenting our process. Another path is to license or to spinoff the technology to another company who wants to take what we do with it in fashion, but in another domain. Someone, or together, could independently use it for other content providers in other markets. The third path would be an evolutionary leap above any and all others in our space to the degree we become an acquisition target — either our brand Runway Manhattan, or the entire Tryon Media company.
What is the market size?
Seven and a half billion dollars is spent annually on imagery for fashion publishing
Have you previously raised investment capital?
When we realized that what we wanted to do in this industry was beyond our own means, what money we were able to put into the company, we started reaching out to investors. We opened our 2.5m convertible note in 2014 and have raised 2 million of that round so far. Of our dozen investors, eight of them are from Duluth, Minnesota.
How many employees do you have and where are they located?
We have five people on the team here in Duluth.
How is the company doing, financially?
We continue to grow our overall revenues, and are close to a break-even point, and profitability.
What is biggest risk or uncertainty in terms of where you are currently vs. where you want to be?
There are lessons learned in the struggle that lead to success. In terms of product-market fit, we have adapted to a moving target. For publishing within the fashion industry, that target started to move slowly about3-4 years ago andis now rapidly changing in terms of how we create our digital content, how publishers use it, and consumers interact with it.
How soon are the industry clients going to embrace a service like Fashion Brain? – that’s the big question.