From Research Papers to Revenue: How Chowon Built Her First AI Products

My name is Chowon Lee, my English name is Chloe, and I’m based in South Korea. I studied Psychology at university, which I entered at 13, and after graduating in August 2025 I fully committed to building products.

Right now I’m working on three. GetThis helps users manage tasks through voice input and screenshots. PostPolish is a Chrome extension I built because as a non-native English speaker I struggled to engage on social media in English, so it generates and translates replies in-context directly inside the interface. Psycled is my first product, still in development, a journaling app for people who struggle to recognise and understand their own emotions.

All three came directly from problems I or people around me actually experienced. That is still the starting point I would recommend to anyone.

My original plan was law, because I wanted a stable career and thought studying human behaviour would be useful for it. Once I was actually studying psychology, I realised it fit me far more naturally. The interest in emotions, cognition, and how people make decisions gradually expanded into wanting to build things that addressed those areas.

The real shift came when I became the leader of a team building a mood-tracking app. That was when I understood that product building genuinely suited me and felt exciting in a way that academic research alone did not.

After graduating I made the decision to fully commit to building. It was both a freeing and an anxious moment: one phase of life had ended, and watching friends choose more stable traditional paths made the uncertainty feel very real.

My first real customer came in January 2026, after I launched on Product Hunt. The product ranked number three Product of the Day, which gave it a significant amount of exposure very quickly.

Because of that visibility, I got my first paying customer only a few days after the launch, and that moment was what pushed me to fully commit to the business.

Before Product Hunt, I had spent months finding early users the harder way: reaching out individually through Reddit, Discord, and X to people who might genuinely relate to the problem I was solving. Some ignored me, some rejected me, some got me banned from communities for promoting my product.

But occasionally one person would become a real user and recommend it to someone else. Those small interactions shaped the product more than any analytics could.

It is genuinely difficult to find real customers rather than just supportive friends or acquaintances. I had to define the customer persona, figure out where those people existed online, and keep reaching out repeatedly.

In the beginning I could not rely on ads or virality, so I went into communities where my potential users already were and tried to start real conversations about the problem I was solving.

The rejections were frequent. Posts got removed. Accounts got flagged. But the users who did come through that process gave feedback that shaped the product in ways no dashboard could have.

Early users are not just customers. They are collaborators. Treating them that way from the start is one of the most important things I did.

Spending money on paid advertising before I had a clear picture of who my users actually were. I thought running ads would accelerate growth, but before finding product-market fit, I had no real understanding of my target audience.

Performance marketing requires defining your audience precisely from the start, and I simply did not have that clarity yet. The ads ran, the budget was spent, and almost nothing came from it.

Looking back, it honestly felt like donating money to Meta and Google. I should have waited until I had a much clearer understanding of my users before touching paid acquisition.

The better investment at that stage was the time I spent in communities, in direct conversations, in individual outreach.

Pivoting too quickly based on a single person’s opinion. With Psycled, my first product, I made countless pivots including a shift toward B2B at one point. When I look back, most of the people whose suggestions drove those changes were not paying customers. They were advisors, well-meaning but not experiencing the problem firsthand.

A product that tries to satisfy every suggestion slowly loses its identity, and I experienced this directly. It cost months of development time.

If a paying customer who genuinely lives with the problem you are solving tells you something is broken, that is worth acting on. If someone without skin in the game suggests a new direction, that is worth listening to once and then setting aside.

AI is involved in more than 90% of what I do. The tool I use most is Codex, primarily for coding but also for writing, market research, and many other tasks. I also use Gemini, ChatGPT, and Claude regularly.

For image generation I used to rely heavily on Nano Banana, but recently I have been moving toward ChatGPT as the performance has improved significantly.

The honest answer about tools is that what I use changes almost every week. The underlying skill is not knowing one tool deeply. It is knowing how to evaluate and adopt new ones quickly.

For me, OpenAI’s products fit my workflow and way of thinking especially well. But that is personal. Someone else might find a completely different stack works better for them. The important thing is to stay curious and keep testing rather than locking into one setup too early.

The two dimensions that matter most to me are time and relationships. Being able to work according to my own routine and physical condition gives me a sense of stability that a fixed schedule never did. And being able to choose who I collaborate with is a genuine source of happiness.

My father and I actually formed a kind of co-founding relationship. He helped with parts of my products and I helped with some of his work. That kind of collaboration, built on trust and genuine fit, is something I could not have had in a more conventional path.

Making new decisions every single day can be exhausting. But it is also what I enjoy most. There is always new information, new trends, new ideas to absorb, and being able to continuously reshape the direction of my own business based on what I learn gives me a strong sense of forward motion.

Everyone naturally wants to avoid risk, and calculating risk carefully before decisions is always right. But I have thought a lot about this: sometimes not taking action increases your risk rather than reduces it.

Because society changes so quickly now, staying still and relying on a stable traditional path can itself become a risk over time. Constantly adapting and pursuing change may ironically be one of the best ways to reduce long-term exposure.

That mindset helped me become more willing to try new things. There were moments when the path I was choosing looked more dangerous from the outside. But taking the risk of building something felt safer to me than the risk of not trying at all.

Start by solving your own problems. Look at your own expertise, your own field, your own daily frustrations. The best starting point is almost always something you genuinely need yourself and that people around you also need.

It does not have to be a huge or revolutionary product. Sometimes simply offering a small service to one person can become the beginning of everything.

And do not focus too much on the distant future. Start by spending just one hour in an evening moving toward what you want to build. What matters most is continuing to move forward little by little. Slow progress is still progress, and it compounds.