What does it take for a small team—or even a solo founder—to build software products and attract customers around the world from day one?
Chowon Lee, a 19-year-old founder from Korea, entered university at age 13 and graduated with a degree in psychology at 18, ranking first in her class. Today, she is building three AI products simultaneously. Her task management app, GetThis, reached #3 Product of the Day on Product Hunt.
Although still at an early stage of her entrepreneurial journey, Chowon has approached product development with a global mindset from the beginning. We spoke with her about building for international users as a non-native English speaker, what psychology taught her about designing AI products, and what small and mid-sized businesses should consider before expanding overseas.
Chowon Lee Co-Founder, GetThis / CEO, Psycled Seoul, Korea Chowon graduated from Chonnam National University with a degree in psychology, ranking first in her class (GPA 4.35/4.5). She is currently developing three AI products: GetThis (voice and screenshot-based task management), PostPolish (an AI-powered writing assistant for social media), and Psycled (AI-driven emotion tracking and mental health analysis).
While she previously led a seven-person product team, she now operates largely as a solo founder, working alongside her father, who serves as co-founder and CTO.
Building for Global Users from the Start
— You started building products for a global audience before you even had your first paying customer. Why?
I never really thought of global expansion as a separate strategy.
When I started exploring SaaS, most of the products I encountered were already serving users around the world. That was simply the environment I learned from. As a result, I never saw Korea as the only market for a product.
My first target was the United States because it has one of the most established SaaS ecosystems and a large community of early adopters. When GetThis got its first paying customer from the US, it reinforced my belief that even a small team in Korea could build products for a global audience.
Looking back, I think many founders wait until later to think about international users. But decisions around language, onboarding, and distribution are often made much earlier than people realize. That's why I believe it's worth considering global users from the beginning, even if your company is still small.
Beyond Translation: The Communication Challenge
— PostPolish was born from your own frustration writing in English on LinkedIn and X. What was actually hard about it?
Writing grammatically correct English wasn't the hardest part. The harder part was making it sound natural for the platform and audience.
A LinkedIn post feels very different from a post on X. Even replying to someone can be tricky. You want to sound professional and confident, but not overly formal or aggressive. As a non-native English speaker, I often found myself switching back and forth between social media and AI tools, rewriting the same sentence multiple times before posting it.
I noticed that many other founders were doing the same thing. They had valuable ideas to share, but the process of translating, editing, and refining their writing created a lot of friction.
That's what led to PostPolish. Instead of opening a separate chatbot and copying text back and forth, users can write directly in the browser and improve their content where they're already working. The goal wasn't just better English. It was making communication feel more seamless.
One thing I learned while building the product is that many non-native speakers already know what they want to say. Often, they're just unsure whether their message sounds natural enough for a global audience. That hesitation can prevent people from sharing ideas they would otherwise contribute.
I think this is particularly relevant for companies expanding overseas. In many cases, the challenge isn't a lack of expertise or product quality. It's finding a way to communicate that value clearly and confidently to people in different markets.
Building AI Through a Psychology Lens
— Psycled is your most psychologically complex product. How does your academic background actually influence the design?
My academic background has probably influenced Psycled more than any other product I've worked on.
As a psychology student, I spent a lot of time thinking about how difficult it is to understand people accurately. Research teaches you to be careful about the conclusions you draw from data and to recognize that human behavior is often more complicated than it first appears.
That perspective influenced an important decision I made while designing Psycled. Instead of building the product around ongoing conversations between the AI and the user, I chose to focus on structured, one-time feedback.
The reason isn't that I think conversational AI has no place in mental health. In fact, there's still active discussion about where these systems can be helpful, where their limitations are, and how they should be used responsibly. As the technology continues to improve, we may develop a much clearer understanding of its role.
For now, I felt it was important to build around a use case with clearer boundaries. Psycled allows users to submit journal entries, reflections, or freeform thoughts, and then receive a structured analysis of the emotions and themes contained in their writing. The goal is not to replace human support or act as a therapist, but to help people better understand their own emotional patterns.
More broadly, I think one of the most important questions in AI product design is not just what AI can do, but what it should be responsible for. Psychology didn't make me skeptical of AI. If anything, it made me more aware of the importance of defining those boundaries carefully.
Building Reliable AI Products
— What was the hardest technical challenge you faced building AI-powered products?
One of the biggest challenges was getting structured and consistent outputs from large language models.
When people first start working with AI, it's easy to assume that once you connect a model to your application, the outputs will be predictable. In practice, that isn't always the case. Large language models are designed to generate natural language, not necessarily the clean, standardized data that software systems often need.
Early on, we saw outputs change based on surprisingly small differences in prompts. A single word, a punctuation mark, or a change in instruction order could sometimes affect the result.
Over time, we found a few ways to improve reliability. Using more capable models helped, but just as important was building better validation and processing layers around the model itself. As tooling matured—including structured outputs, tool-calling frameworks, and standards such as MCP—it became easier to create more reliable workflows between AI models and external systems.
One lesson I took away from that experience is that deploying AI is often more complicated than it appears from the outside. The model is only one part of the system. Prompt design, output validation, monitoring, and ongoing iteration all play important roles in making sure the final user experience remains reliable.
That's especially important because models continue to evolve over time. A workflow that performs well today may behave differently after future model updates, which means quality control can't be treated as a one-time task.
Toward Preventive Mental Health Technology
— What are you building toward long-term?
In the long run, I'm interested in building products that go beyond software alone.
One thing that has always fascinated me is how often people recognize burnout, chronic stress, or anxiety only after they've already reached a difficult point. Looking back, many people can identify warning signs. The challenge is that those signals often go unnoticed in the moment.
That's a question I've been thinking about for years: what if those signals could be detected earlier?
I believe the future of mental health technology lies at the intersection of psychology, AI, and physiological data. Heart rate patterns, sleep disruption, stress-related indicators, and other biometric signals may contain information about our mental state that we're not consciously aware of.
My long-term vision is to build a wrist-worn device focused specifically on mental well-being—something that could help identify early signs of burnout, chronic stress, or panic-related symptoms before they become overwhelming. Instead of helping people understand what happened after the fact, I want to help them notice what's happening while it's still possible to respond.
In the short term, I'm focused on continuing to build AI products, raising investment, and learning from startup ecosystems such as San Francisco and New York.
I've always been interested in understanding people. Psychology led me to study people. Building products became a way to help them.
What Japanese Businesses Can Learn
Although Chowon's products are still at an early stage, her experience highlights several challenges that many small and mid-sized companies face when entering international markets.
Building for global users requires more than translation. It affects product design, communication, customer acquisition, and even the assumptions made during the earliest stages of development. Throughout the interview, Chowon repeatedly returned to the same idea: decisions made early tend to shape everything that follows.
Her experience also suggests that international opportunities can emerge sooner than many companies expect. For Chowon, some of the earliest signs of product-market fit came from users outside Korea, reinforcing the importance of paying attention to where genuine demand originates.
As Japanese companies continue to explore overseas growth opportunities, the barriers to reaching global customers are lower than ever before. At the same time, success increasingly depends on a company's ability to communicate clearly, adapt quickly, and build products with international users in mind from the beginning.
Key Takeaways
- International users may appear earlier than expected. Pay attention to where your first customers come from.
- Expanding overseas is not only a localization challenge. Communication, positioning, and trust matter just as much.
- AI can create significant value, but reliable deployment requires clear boundaries, validation, and ongoing monitoring.
Readers interested in learning more about Chowon’s work or exploring potential collaborations can connect with her through the channels below.
- GetThis: getthis.day
- Psycled: psycled.com
- PostPolish: social-reply-polisher.vercel.app
- LinkedIn: linkedin.com/in/chowon-lee
- Email: chowon.m.lee@gmail.com
Leap supports Japanese small and mid-sized businesses in building multilingual web presence and executing overseas digital marketing. If you're considering your first step into global markets — or looking to strengthen your existing presence — we'd be glad to talk.
Contact Leap: https://www.leap.site/en/about/#contact
