After graduating from UT Austin in 2010 with a master’s degree in electrical and computer engineering, Sowmiya Narayanan went on to co-found Lily AI — an AI-based software designed to enhance users’ online shopping experiences — and she now serves as its chief technology officer.

We sat down with Narayanan to hear about her time as a graduate student at UT, her life as an entrepreneur running a successful and growing technology company and her thoughts on the future of artificial intelligence.

Sowmiya Narayanan, Cockrell School of Engineering alumnus

What impact has your education at UT had on your career?

While at UT, I had the opportunity to deeply dive into what I wanted to do. The professors and advisors were ready to help me succeed, and I had a wealth of opportunities at my fingertips to really immerse myself in my work and research. As a TA, I was also able to connect with a lot of undergraduate students whose perspectives and insights enriched my education. Post-graduation, I have had the opportunity to tap into UT’s powerful alumni network, which has been invaluable as I navigate the entrepreneurial landscape.

How does Lily AI work?

Lily works on the backend of a user’s shopping experience. While you are browsing websites and shopping as you normally would, Lily is following your behavioral patterns and analyzing your emotions and purchasing motivations. Our team then works with brands and retailers to gather additional data like product descriptions and basic user profile information before we start building a customized profile for the user. As you continue to shop your favorite websites, you will begin to see more individualized product suggestions, resulting in a better online retail shopping experience. You, as the user, aren’t doing anything different, but your experience will be increasingly different and more customized.

What are some of Lily AI’s goals for the next 5-10 years?

We want to develop a very deep understanding of both products and users. When we are able to tie these two pools of information together, we will be able to significantly increase retail function. Today, we are simply doing product recommendations, but we hope to one day be able to forecast purchasing decisions so we can help users make the best buying decision. We’d also love to expand into other domains like home and beauty. Our end goal is to have one backend profile for a user through which she or he can rely on Lily’s ideas to meet any retail needs.

You started Lily AI with your business partner, Purva Gupta. What advice do you have for solo entrepreneurs who may be seeking a business partner?

First and foremost, it is critical that you and your business partner(s) believe in the problem you are trying to solve. You both have to be on board. Purva and I share that. Second, you need to believe in your ability to solve that problem. Sometimes you have to break it down into smaller, more manageable steps, but your perseverance — and that of your partner — will be crucial to your success.

Purva and I also complement each other’s skill sets — she comes from the business side, and I come from the tech side. We are stronger together than if we were tackling this problem alone, and we are both passionate about focusing our company around key core beliefs. For us, we want to empower women. Lily AI revolves around shifting the relationship between women and the fashion industry from a narrative of body shaming to body positivity. As a team, we believe in this mission and, therefore, everything we do and the product we produce radiates those same values.

What changes do you foresee in the fashion industry as AI continues to advance?

So far, most product personalization has been done on segments of users. Unfortunately, many companies that have tried this segmented approach have flatlined because there are limitations to how much you can personalize a given segment. Two people are naturally going to have different feelings, emotions and motivations for buying the same item. However, true personalization will revolutionize this sphere. When we can hone in on traits like a person’s emotional context, their aspirations, their body image, etc., we will create a highly individualized approach to shopping. Eventually, this will eliminate the doubt a user may have when viewing products that aren’t right for them — when you are viewing products catered to you, there will be no room for body shaming, self-deprecation or the negative self-talk often associated with mass product recommendations.