Alumnus Dae-you Kim enrolled at Seoul National University in 2021 and is currently majoring in business administration and computer science at the College of Liberal Studies. He is also the CEO of Iruri Labs. Kim won the top prize at the 2026 Startup Convention (Entrepreneurship Competition) for PassDream AI and now offers a range of services, from workbook analysis to automated grading of student answer sheets.

1. Please introduce Iruri Labs and your responsibilities as the CEO.

I am Dae-you Kim, the CEO of Iruri Labs. Iruri Labs is a startup that uses AI to automate grading and learning management tasks for lecturers at private academies. If a lecturer uploads a PDF or images of a workbook at once, our AI analyzes the questions and makes them gradable. After that, when students send photos of their answer sheets, they can be automatically graded.

I am a third-year student at the College of Liberal Studies, majoring in computer science and business management. I am constantly studying or working and actually came running here after a lecture today (laugh). I am always running out of time, but the campus itself has become a resource for me. There is the SNU Entrepreneurship Program, and our teammates are mostly undergraduate students at SNU. Being based at SNU has become a brand for us, and we are also benefiting from its network.

2. What are the features unique to Iruri Labs?

Our most notable feature is our universal grading system. Users can grade all types of questions, including essay, short-answer, and multiple-choice questions, from elementary, middle, and high school curricula, regardless of the subject (such as Korean, English, math, social sciences, or science) or the test format. We do not support arts or physical education subjects, but we do support nearly all types of test papers used in private academies.

We also provide various follow-up functions after grading, such as a clinic feature that collects all questions with incorrect answers and automatically creates a test paper, as well as integrated features ranging from learning statistics analysis to automatic notifications for parents. With a focus on grading, we have automated the entire post-assessment process.

Existing automated grading services in the market can usually grade questions created on their company's platform using OMR cards or tablet-based touch methods. We are the only company that can grade any workbook immediately, whether it is uploaded as an image or a PDF.

We also do not use OCR. Our service combines our VLM-based pipeline with our in-house models and orchestrator, enabling us to process even handwritten text and essay-style answers.

Unlike commonly known problem-solving services such as QANDA, which simply search for pre-stored answers, our service is fundamentally different because our proprietary model thinks in real time and grades the answers itself.

We have also lowered the cost burden thanks to our efficient structure, which can process up to 15,000 questions for 75,000 won a month.

3. What inspired you to start your own business?

I first wanted to start my own business in middle school. I graduated from a small alternative boarding school in Ulsan with only 60 students. The school invited various social entrepreneurs for special lectures, and that is when I met a CEO who grew and distributed lettuce.

I still remember what the CEO said. Lettuce is usually cheap in the summer because it is so common, whereas its price is ten times higher in the winter, when it is harder to farm. The CEO started the business with the principle that the price difference between the two seasons would not exceed double. At first, the company had 100 clients, but in the summer 90 left, and only 10 remained, who told the CEO, “When winter comes around, we will have to buy your lettuce anyway.”

What the CEO emphasized was that a company does not become a social enterprise by earning a mark, but that not deceiving people and being kind in business can have a significant social impact. That comment resonated with me. After that, I came to think that startups have the potential to have the most widespread, fastest, and positive social impact, which became my most fundamental motivation for starting my business.

4. Please walk us through the process of establishing Iruri Labs.

We became a corporation in May 2024. My co-founder served in the same unit as I did during my time as a soldier, and he was a personnel officer for 7 years. He has experience establishing a new administrative system in the division, so at Iruri Labs, where we are working to change the administrative systems in private academies, he is fully using that experience to lead overall planning and sales.

Shortly after establishing the corporation, we launched a business offering test preparation for professional licensing exams. We also provided mentoring by pairing test-takers preparing for CPA and CTA exams with practicing professionals. We sold digital textbooks and recruited more than 20 mentors for the program. We then met a web developer at the Like Lion cohort, and the three of us formed a team. We still maintain that app, receive payments from test-takers, and generate about 1 million won in revenue each month.

However, we did not see any growth in sales even after 7-8 months. A PhD researcher studying startups, whom I met through an acquaintance, gave me some harsh feedback, saying, “You do not have the figures, the market is small, and the growth potential is limited. This is not a startup model.” I took that feedback seriously and decided to pivot.

We found a direction for our pivot through our existing networks. We spoke with lecturers at CPA and CTA academies who told us, “The second exam is essay-based, so grading it is expensive. Private academies would be willing to pay for it if the grading can be automated.” This insight led us to develop an AI grading solution. We initially tried to build a grading tool tailored to CPA and CTA exams, but as we developed it, we found there are only about a dozen private academies specializing in CPA and CTA exams. We shifted our focus to a general-purpose grading solution, which is now Iruri Lab’s core service.

5. Did you experience any challenges when you launched your business, and if so, how did you overcome them?

Launching a startup always involves a continuous cycle of slumps and death valleys, but the biggest challenge came just before we pivoted. We did not generate much revenue after 7 to 8 months of operating at full speed, and we received negative external feedback. At that time, our developer could no longer work with us for personal reasons and left to join a large corporation, leaving the co-founder and me as the only ones left.

However, no one left the team on bad terms. There has always been a culture of respecting each other’s feedback, and I also feel that we have been fortunate. Pivoting is not always emotionally easy, but persistence has never been my forte. Instead, I tend to view situations objectively. I think that if something does not seem to work out, you should call it quits, which I believe helped me make the decision more quickly.

Another hurdle came in the summer of 2025, when we were in our development phase. We had three developers, including me, and we had to design the grading pipeline structure from scratch. Determining whether a question was answered correctly or incorrectly proved far more complex than we had expected, so we had quite a few arguments, but in the end, it all paid off in August, when we first confirmed it was working.

I think responsibility is what allows me to withstand everything. I always think I should not give up despite the challenges, because I have co-founders and employees who trusted me enough to join the company.

6. What would be the most memorable moments after launching Iruri Labs?

I remember two moments.

The first is from the early days when we were running the mentoring platform. We received several comments from test-takers who passed the CPA and CTA exams through our mentoring service, and it was rewarding every time. Although we did not make much profit, that feeling is still a good memory for me.

The second memory is from the summer of 2025. It was extremely difficult to develop the grading pipeline, but we were motivated and worked through the night after an investment analyst I met through an acquaintance told us to “Get the product ready by August.” I, for one, wrote about 200,000 lines of code on my own and built five AI models in just two months. In the first week of August, we entered the questions and graded them, and it worked. The feeling I had when I watched that moment with our developers still lives fresh in my mind. The challenges made it even more memorable.

7. Please share Iruri Labs’ pivot story.

We had two major pivots.

First, we pivoted from providing test preparation services (mentoring and digital textbooks) for professional licensing exams to AI grading solutions for CPA and CTA exams. The mentoring platform generated virtually no revenue for 7 to 8 months, and given the harsh external feedback, we decided to pivot. That was when lecturers at CPA and CTA academies told us, “Grading essay-based tests is so expensive. It would be a lifesaver if the grading could be automated,” and we decided to shift our direction.

The second pivot was our transition from offering grading for CPA and CTA exams to a general-purpose grading solution. As we developed the technology, we found that there are only about a dozen private academies for CPA and CTA exams, which means that selling our solution to those academies would limit our market. So, we thought, “It might be a waste to limit this good technology to a smaller market,” and redirected our focus to developing a universal solution that grades all subjects across elementary, middle, and high school. That is the current “PassDream” solution.

At every pivotal moment, what I valued most was the bond I shared with my teammates. We have consistently emphasized that we are not merely changing to survive but expanding to create greater value. We have held discussions to reach consensus and, thanks to this integrity, could move to the next stage without wavering. In retrospect, the pivoting process was a meaningful journey for the team as we worked together to find the right answers for the market.

8. What are your visions for the company and for yourself?

To be frank, I want to make more money and be recognized as a successful entrepreneur on a personal level. As the CEO of the lettuce farm told me when I was in middle school, I have always believed that simply running a business without negatively impacting society is, in itself, a significant impact. These two are the foundations of my motivation.

Iruri Lab’s vision can be broadly divided into two tracks.

The first track is to go beyond automated grading in the EdTech field and fully automate the post-assessment process (analysis, report generation, and integration with complementary learning). We are the only company providing a general-purpose solution in Korea. 80 private academies across the country currently subscribe to our service, and we plan to expand our client base to over 250. We are also considering overseas expansion, starting with China and Japan, which have educational environments similar to ours. China has significant market potential because it has academies with 620,000 students at their headquarters.

The second track concerns technology. We collect more than 10,000 grading records daily and have collected 650,000 records over four months since the official service launch. For grading, it is pivotal that the AI is reliable and free of hallucinations. We are developing an engine called “Atlas” to improve reliability, and I believe that once it is complete, it will be a technology that can be sold not only to EdTech companies but also to large corporations and LLM providers.

9. Please share the insights you gained about your target consumers while launching your company.

First, I thought our product would mainly be used by younger lecturers who are early adopters. However, I found that a considerable number of older lecturers who are not familiar with computers are using our solution. Some of them ask us how to edit PDF files or create shortcuts to the website on their desktops. The only reason they want to use our service is to be free from grading. The desire to be free from grading tasks was the same among all teachers, regardless of age.

Another interesting discovery was how creatively our clients could use our product. One private academy limits phone use, so they have set up an unused phone in the lobby for students to take photos of their answers as soon as they arrive. It was not a use we had considered in our designs, but it gives us hints about which features we need to develop in the future.

In terms of student composition, our main clients are middle school students in their first and second years. Academies for high school seniors found it difficult to allocate resources to implement a new system, and we found that handwriting recognition accuracy had not yet been perfected for lower-grade elementary students, leading to a higher omission rate.

10. Advice for Junior Students

I first want to tell you that I am not a successful person; I am a challenger. So, what I share is definitely not the formula for success.

Nevertheless, I have one piece of advice to give you: if you want to do something, do it.

However, how you start doing something is important. Of course, you should not jump into something without any preparation. I thought I was well prepared, but I still experienced failure. But if you set your eyes on something, prepare, and start doing it, you will have to take responsibility even when things become difficult, because it is a path you yourself have chosen, which kept me going.

Challenges leave you with something. You constantly think about things even when you shower, and in the end, you gain something in the process. And if you are really successful, you can have a positive impact on our society and the people around us. If tests are graded well, it makes things easier for instructors and gives students better, faster feedback, which helps them get closer to their goals.

In the end, it is the challenger who creates something.

The Stories from the Entrepreneurship Center section features interviews with current and former students majoring in Entrepreneurship, as well as startup teams that participated in the Center’s programs, sharing the journeys of young entrepreneurs. If you are interested in this section, please email snustartup@snu.ac.kr.