©2026 Business Next Media Corp. All Rights Reserved. No.102, Guangfu S. Rd., Da'an Dist., Taipei City 106, Taiwan
In 2018, Yusan Lin was just another PhD student in Palo Alto, California, working on her dissertation at Café Venetia. One day, a professor sitting at the next table turned to her and asked about her research. Yusan Lin explained that she was working on using AI to detect fashion trends. The professor listened, then handed her his business card and said she could reach out if she ever needed help.
It was only later, after a quick Google search, that Yusan Lin realized who she had met—David Cheriton, Google’s very first investor. “Even now, I still can’t believe I actually met David Cheriton,” says Yusan Lin, founder of Mirror Mirror AI.
That chance encounter made a lasting impression on Yusan Lin. Living in Silicon Valley, the world’s startup epicenter, she was used to overhearing investment discussions and startup pitches in Mountain View cafés. For the longest time, she hadn’t paid much attention to these conversations, but meeting David Cheriton made her realize she hadn’t fully taken advantage of her unique surroundings. “It feels like a missed opportunity,” she reflects.
Seven years later, things have come full circle. Yusan Lin’s AI fashion startup, Mirror Mirror AI, secured a $1 million investment through the a16z Speedrun program.
As a PhD student in computer science at the University of Pennsylvania, it might seem logical for Yusan Lin to launch a startup based on AI. But why combine artificial intelligence with the fashion industry? The answer goes back to another one of Yusan Lin’s dreams: becoming a model.
From a young age, Yusan Lin was fascinated by both fashion magazines and coding. In college, she got the unexpected chance to walk in a campus fashion show. “That experience opened up a whole new world for me,” laughs Yusan Lin. She even passed the casting for Catwalk, one of Taiwan’s top modeling agencies. “My parents only found out I was modeling when they saw my picture in the newspaper. They wanted me to quit, so eventually I called to withdraw from the agency. But I still did some runway shows from time to time, spending a lot of hours on the train between Taipei and Zhongli.”
That unexpected opportunity made Yusan Lin start thinking seriously about life after graduation. Should she pursue a modeling career, continue in academia and aim to become a professor, or follow her classmates into the tech industry as an engineer?
In the end, her decision was a bit unexpected: she chose to become a university professor. “At 22, you’re still figuring things out. I didn’t want to work as an engineer in a tech park, and if modeling didn’t work out, I wanted to keep studying,” says Yusan Lin.
After winning a full scholarship to the University of Pennsylvania, Yusan Lin became an engineering PhD student, a path that eventually brought her to the cutting edge of the tech industry. She worked as a research scientist at Visa, building fashion recommendation systems, and later joined Amazon Fashion as an AI scientist, focusing on size recommendations and virtual try-on features.
“At first, I was really excited to get offers from these big companies. When I went out for coffee, people all around me were pitching ideas and talking about investments, but I didn’t pay much attention to it,” she recalls. “Then I read ‘Bad Blood,’ the book about Theranos, and realized that Elizabeth Holmes used to go to the same cafés I did. Suddenly, it felt like I was living in a parallel universe.”
That was the moment that set the stage for Yusan Lin’s now-famous encounter with David Cheriton and sparked her entrepreneurial ambitions. “Maybe it’s just being in Silicon Valley—this environment really rubs off on you.”
The idea for Yusan Lin’s startup first took shape at the end of 2023, when generative AI was taking the tech world by storm. She figured that if AI could generate images, then fashion should become something accessible to everyone.
The first version of Mirror Mirror AI was an online personalized fashion magazine app. Users could open the app and see themselves as the model. The app would recommend outfits to buy, mix and match them with pieces users already owned, and help inspire new styles. Inspired by this idea, Yusan Lin decided to quit her job and officially launch the company in May 2024.
But this romantic vision quickly ran up against the harsh realities of the market.
“I only worked on that idea for three months,” she says. The biggest setbacks came from two areas: technology and investors. “Back then, in the summer of 2023, AI image generation was completely different from what it is now. We brought in students from the Fashion Institute of Technology in New York to test it out, but the results were awful. I even had to apologize to our users for the poor images.”
Investors were also a challenge, as they had little interest in Mirror Mirror AI’s original concept. Looking back, she admits that in the early days, she was too focused on what she wanted to build, instead of considering what the market actually needed. She also made a classic rookie mistake: only asking friends and family for feedback, which led her to misread the market. “My advice is, never ask friends and family. They’ll always blindly support you,” she laughs.
These setbacks taught her a tough lesson: building in AI is a game that requires capital and resources. If a small team can’t create beautiful, high-quality images with their own models, they simply can’t compete.
With the consumer route temporarily off the table, Mirror Mirror AI pivoted to the business market, where demand seemed stronger. The company began offering AI solutions to fashion brands, such as generating images for social media, e-commerce, and flat lay photos.
This pivot gave her a much deeper understanding of industry pain points and the real process of achieving product-market fit. She observed that tech-driven teams often try to create dazzling features, but the actual needs in fashion are much simpler. Most brands just want to cut costs or move faster. Even startups’ visions of “cost reduction” can be overly romantic.
“For example, I visited a major fashion brand’s studio in Los Angeles, which basically runs like a massive factory. They have photographers, models, and makeup artists all in-house. From the moment a model steps in front of the camera to the time the product is live on their website, it only takes 24 to 48 hours. It’s extremely efficient and low cost—cheaper than what most AI startups could offer. That’s why a lot of big brands think it’s faster to do things themselves,” says Yusan Lin.
In other words, for AI startups hoping to break into the fashion industry today, the real competitive edge isn’t just about technology. It’s about truly understanding the vertical industry.
Yusan Lin believes that with AI models evolving so rapidly, it’s a huge waste for small companies to pour resources into training or fine-tuning core models. She says the real advantage should be in UI and UX, making the product seamlessly fit into the customer’s workflow and actually improving day-to-day productivity.
But at this stage, Mirror Mirror AI ran into another classic startup pitfall: becoming a custom solutions shop. “When a big brand is willing to talk to you, it’s so exciting and you want to build anything for them,” she recalls. It wasn’t until later, after realizing that most of those features didn’t matter to other clients, that the team put on the brakes.
She offers a survival guide for entrepreneurs building in the age of AI. First, find the common pain points among your customers. Second, always keep up with the latest developments in AI models and sharpen your judgment. “A simple rule is this: if Google is already working on something, don’t put your own resources into it.”
Of course, shifting a business model still requires capital. In April 2025, Mirror Mirror AI applied for the a16z Speedrun program. This is an accelerator-style incubator inside Andreessen Horowitz that invests $1 million and provides three months of mentorship to early-stage teams.
“We didn’t get in this time! But honestly,” she laughs, “it was the best rejection letter I’ve ever received.”
What made it so great wasn’t just that it was from a16z. The letter included specific, constructive feedback. They praised her background but pointed out some real gaps: Mirror Mirror AI needed stronger pricing strategies and business insights. They suggested finding a co-founder with deeper experience in fashion. “World-class VCs really are on another level,” she says.
Those reflections prompted her to pivot again. By August, she shifted the product focus once more. This time, Mirror Mirror AI attracted angel investment from industry leaders like Microsoft product chief Aparna Chennapragada and Thinking Machines CTO Barret Zoph. A CTO friend at Zynga even encouraged her to reapply to Speedrun, saying the a16z team wanted to see her try again.
She moved quickly and seized the opportunity. From reapplying to receiving a term sheet, the entire process took just two weeks. Mirror Mirror AI secured investment from a16z. “I asked them why they came back to us, and they just said they thought I was the right person for this idea.”
So, what’s the latest product from Mirror Mirror AI? For now, Yusan Lin gives a classic Silicon Valley answer: stealth mode. The company isn’t revealing details yet.
“What I can say is that there are a lot of debates about AI-generated images in the fashion industry. I don’t think the industry is ready for pure generative tech just yet. That’s why our third pivot is toward building infrastructure, more of a platform for fashion,” she says. Stealth mode lets her team focus on product and market fit without premature exposure, especially when technology changes so quickly. If you announce your positioning too early, you might need to change again by the next quarter. “If I can raise from top VCs while in stealth, then when we finally announce, it creates FOMO and more investors want in,” she says with a smile.
Now that Mirror Mirror AI has secured investment from a16z, the next step is all about refining the product. And what about her modeling dream? “Actually, I ended up signing with a small agency,” she laughs. “I finally got signed! But honestly, I never had time to do any actual modeling. Starting a company keeps you way too busy. I guess I’ll just count it as a badge for a dream fulfilled.”
©2026 Business Next Media Corp. All Rights Reserved. No.102, Guangfu S. Rd., Da'an Dist., Taipei City 106, Taiwan