[NELog] OD Lee Seonjun 'A Special AI That Will Vertically Boost Your Appearance Favorability Index'

Editor's Note[NE Log] is a video content series produced for companies selected in ‘NEXT ELEVATION,’ a startup accelerator program established by Asia Economy.

Top U.S. University Engineering Student Becomes AI Fashion Platform Developer

Hello, I am Lee Seonjun, CEO of OD. I majored in a combined program of Computer Engineering and Mechanical Engineering at UC Berkeley in the United States. During that time, I studied machine learning extensively, and even after graduating, I continued researching and developing applications in machine learning to build my expertise.

I have always been interested in fashion. Since I was young, I liked clothes and enjoyed dressing well. Recently, I sensed several times that trends were changing rapidly. I noticed that people no longer want big, famous brands but prefer unique, small, promising new brands. I was the same. I spent a lot of time searching for those small brands and unique pieces. Recently, many applications based on LLMs from OpenAI, like ChatGPT and Bard, have become commercialized. Using those, I thought, 'How can I create an AI that understands my taste through LLM technology and find new brands, designers, and producers through that AI?' That idea was the starting point.

LLM

Large Language Model (LLM) refers to artificial intelligence trained on massive text data to understand and generate human language. One of the most famous LLMs is OpenAI's GPT-3 (Generative Pre-trained Transformer 3).

Models based on LLMs like ChatGPT or Google’s are quite common. However, I wanted to create a model composed solely of fashion domain data, so I considered how to approach fashion using technologies from Google or Facebook (now Meta). Early on, I realized foundational technology was necessary. That’s where my machine learning studies at Berkeley helped a lot.

Fashion Platform’s Warring States Period: OD’s Emotion and Value

There are many online platforms like Musinsa or 11st in Korea, and similar ones overseas. I don’t think OD is about competing with those platforms. For example, if I want to find jeans, I would first go to Musinsa or 11st, enter the ‘Men’ category, then ‘Pants,’ and filter through options. This search process has been the same since the internet began and these platforms emerged. I don’t think this user experience has ever changed through technological innovation.

But recently, new user experiences and new LLM technologies have made it possible to create different experiences. If we develop that, it could be OD’s biggest differentiator.

Simply put, when you go to Google.com, there is a search bar where you type what you want, and results appear. Similarly, on OD, there will be a chat bar where you can discover what you want and the style you’re thinking of (for example), 'I’m going to a steampunk-themed party, so I’m looking for leather shorts style' through chatting. That will be the user experience.

AI is a very hot topic these days. Creating items that help a lot in fashion and discovery fits this era well. Consumers and shoppers now value curation the most?personalized curation. They no longer want big brands but small, unique brands and unique clothes, but the common opinion is that ‘finding them is too difficult.’ That’s why I thought we could create a suitable AI solution.

At 20, First Startup... Young Entrepreneur Who Even Declined Google Job

It’s not that I declined; I just postponed it. I also postponed offers from other companies. I was clear about what I wanted to do. Growing up, I saw my parents running a company they started 25 or 26 years ago, and I wanted to start a business from a young age.

I started a business right after entering university with the mindset of 'Let’s study and start a business!' That mindset never changed. When I entered Berkeley in 2016, I launched my first startup called ‘Endorsify’. Back then, terms like creator and influencer were not commonly used. The concept of influencers and creators was very early stage, and how creators could earn money was also in its infancy. At that time, we created a service marketplace that helped deliver influencer marketing effectively.

One of the biggest problems then was that there was no way to measure how valuable a creator’s ability was. For example, if I had 1 million subscribers and a brand wanted to advertise through me, the client had to pay whatever I asked. The brand couldn’t bet on how effective my influence was. So, we built an influencer marketing business that used machine learning and data science to determine a person’s Fair Value, including the Fair Value per post. We operated Endorsify for about 3-4 years before selling it to a larger company that wanted to build an influencer marketing platform. Endorsify onboarded many influencers and brands, and after the sale, I was recognized in ‘Forbes 30 Under 30.’

The second startup was ‘Minti,’ which started in early 2020. We had the network from Endorsify and investors suggested we explore NFTs. However, the timing for Minti was not good. The NFT hype was huge when we started, but rapid growth was followed by a quick decline. We closed the second startup in 2022.

It was a success to some extent but not a big success. It could be seen as a failure, but I believe you learn a lot from failure. These experiences gave me know-how and confidence to build a proper team. The most important lesson from Endorsify and Minti was timing. I realized how crucial timing the market need is before building a product. Sometimes our good products did well, sometimes not, despite surveys and customer research. Through these experiences, I learned the importance of timing and how to achieve a good go-to-market fit. I’m confident to apply that to this project.

OD started from my idea alone. Since the item requires a lot of machine learning technology, I brought in my friend Brian Wang, who studied machine learning with me at Berkeley and has extensive experience in the field, as a co-founder.

Our next important hire is a creative director. We need talent who can build good relationships with various brands. We are considering candidates who studied at the Fashion Institute of Technology in the U.S. and have steady experience in the fashion industry. We are looking for people who can help develop this new business.

Different Startup Culture in the U.S., Yet Why Start in Korea

I feel more comfortable in the U.S. I consider it my home, and my network and friends are mostly there. So, I am thinking about returning to the U.S. someday. However, I chose Korea as the testbed for OD because the online e-commerce market in Asia is much larger than in the U.S., especially in fashion. No place leads Asian fashion more than Korea. So, we set our goal to start in Korea.

Starting a business is very easy in the U.S. Anyone can start a business, and the support system is well established. The government, states, and companies all support startups well. However, Korea’s startup ecosystem and systems for entrepreneurs are not as developed, especially regarding regulations. In that sense, it is very encouraging to see places like NEXT ELEVATION with a clear mission to build a better ecosystem through new accelerators.

OD CEO Lee Seonjun’s Vision for the Future

I believe everyone has their own style. Of course, some people don’t invest much in developing their style due to lack of time or other reasons, but even those people have the potential to develop their own style and preferences. I hope OD becomes a product that awakens that potential. In that sense, OD’s target is ‘everyone.’ But realistically, the initial target will be people who care about style and need AI to help with discovery.

Jeff Bezos once said his dream when starting Amazon was to become an online retailer that sells everything. It started with selling books online and grew into a huge company. That dream of selling everything online became a reality. Similarly, I am starting with clothes, but I believe the platform that enables discovery through recommendations based on LLM, machine learning, and AI will expand beyond clothes to lifestyle. Starting with clothes, then expanding to beauty, accessories, books, magazines, and other categories for discovery. When people ask what lifestyle AI knows my style and preferences best, I believe OD will be the answer.

This content was produced with the assistance of AI translation services.