Problem Solving

If we look at our surroundings from a slightly different perspective, most things can be seen as a process of problem solving. From individual daily choices and relationships to the process of organizations establishing visions and designing structures, the countless decisions and attempts we face are ultimately within a flow of defining problems and finding solutions to them. Viewing the world through the lens of problem solving helps us organize vague situations into language and gauge our next choices. This may sound obvious, but the same applies to our own concerns. From the burden of major exams to anxiety about career paths, to questions about what we will do with our lives—when we define these concerns not as vague emotions but as problems to be solved, and reinterpret our experiences as resources for solving those problems, what we already have and what we still lack become relatively clear. This allows us to view what actions we need to take next in a more structured way.

Amidst the somewhat abstract desire to create impact that changes our surroundings, I had been contemplating between entrepreneurship and graduate school. For me, the perspectives on entrepreneurship learned in the Business Venture and Entrepreneurship class and a project I carried out as an extension of that became an opportunity to glimpse, however faintly, 'what problems can be solved through entrepreneurship' and 'how can such problems be addressed in the form of a business.' This article was written to record the thoughts and learnings from that process.

I still don't know a clear methodology for choosing good ideas or items. I wonder if this is even a problem with a correct answer. This project was merely a prototype; it didn't generate revenue or change the world. However, the process of defining problems, setting hypotheses, and seeing how those hypotheses transformed in reality has clearly left a great meaning for me, and I think it's enough if the trajectory of that change can provide even a small insight to someone.

This article will unfold the story following the order in which the project progressed, while organizing the perspectives learned in class from Professor Jong Hoon Bae along the way. The professor played a key role in establishing the Department of Business Administration in Entrepreneurship, and through the class, he made me see entrepreneurship not as 'the skill of creating a company' but as an approach to dealing with problems. This article focuses on problem definition and validation, which are part of market test, one of the two major stages of commercializing ideas (market test and scalability).

Entrepreneurship / What is an Entrepreneur?

Before developing the content, I want to organize how we should view 'entrepreneurship' and 'entrepreneurs (or entrepreneurship studies)' that run through this article. The entrepreneurship that the professor defined in class was a unique attitude toward solving the problems of our time. This definition was impressive because it made me see entrepreneurship not simply as creating a company or a means of generating profit, but as an approach to dealing with problems.

From this perspective, the purpose of entrepreneurship is closer to verifying whether problem solving actually works, rather than profit itself. When the problem we want to solve is accepted by the market, the result appears in the form of profit. Ultimately, entrepreneurship can be said to be the process of redefining problems that already exist around us but were taken for granted, and creating a market that can solve them.

That's why excellent entrepreneurs design a market model before a business model. Rather than thinking about what more to sell to people who are already consuming, they consider why some people haven't become consumers yet, and what needs to change for them to enter the market. The starting point of entrepreneurship is not optimizing existing markets, but discovering points that existing markets have missed.

However, such attempts are inherently uncertain. It's difficult to clearly measure or know at the start how big the problem we're trying to solve is, or whether it can expand into a market. So entrepreneurship is not a choice that can be completed through calculation, but a process where you can only know the result by taking risks and pushing through to the end.

What Problem to Solve

So how can we decide what problem to solve? Of course, there's no correct answer here either, but through this project, I've come to view ideas and problems from at least the following two perspectives. In the subsequent content, I'll explore more specifically how these criteria changed the project's direction and what choices were actually made.

• Check the system and infrastructure that constitute the problem.
• First identify the resources and capabilities the team has, and select a problem of a size that can be solved within that scope.

Design Thinking and Infrastructure

Our team also repeated the process of ideation and complete revision for nearly two months, and only in the last month were we able to make relatively rapid progress. That's how arduous the process of exploring 'problems that look good' was. At least the various problems or items we explored included:

• Information gap problems arising from the lack of local news media
• Lack of information accessibility due to insufficient promotion of local communities or campus organizations
• Imbalance in cultural infrastructure concentrated in large cities
• Employment gap for college graduates and concentration of highly educated workforce centered in the Seoul metropolitan area
• Lack of business direction setting and mentoring in the youth entrepreneurship process
• Small-portion vegetable sales targeting single-person households
• Café study space platform for using neighborhood cafés like study spaces
• Subscription management fintech for managing and recommending multiple subscription services
• Consultation and information curation targeting information asymmetry in insurance after traffic accidents
• Entry problem for amateur creators in the music market where consumers and producers are clearly separated
• Response solutions for malicious customers, i.e., 'problem customers'
• Mental health counseling app services dealing with depression or burnout

Most were problems we'd heard about at least once, or that someone had likely already tried. And the characteristics these items commonly shared were also clear: they hadn't been concretized as items yet, had been discussed for too long, and were difficult problems to address. So why do these problems keep appearing but rarely get solved? When we view them from the perspective of infrastructure rather than superstructure, a completely different interpretation of problem definition becomes possible.

An example the professor gave in class helps understand this perspective. The problem of university hierarchy is a social issue that appears periodically in Korea. When approached from superstructure—admission systems, excessive competition, or the success-oriented mindset of the parent generation—the problem always becomes personal, complex and solutions become vague. However, viewing it from infrastructure allows a completely different interpretation. Seoul National University is the school with the largest budget invested per member, where the most resources are concentrated. That's why people flock to Seoul National University. From this perspective, this problem transforms from a B2C problem like individual choice or culture to a B2B problem.

We initially chose and developed a vegetable meal kit for single-person households, but from this perspective, we can anticipate that this too is an item with limitations. Approaching from superstructure, i.e., top-down, centers on psychological narratives like the increase in single-person households, the hassle of cooking, and the desire for a healthy meal. In this case, strategy easily concludes with emotional packaging, storytelling, and warm copy. However, we can easily confirm that such attempts have repeatedly failed. Conversely, viewing from infrastructure, i.e., bottom-up, the problem becomes much clearer. Let's think: the reason single-person households don't cook is not willpower but the difficulty of small-quantity purchases and the inefficiency of storage and disposal. Here, single-portion meal kits inevitably have high costs due to small-quantity production and freshness maintenance, and in B2C marketing structures, customer acquisition costs are also critically high. Ultimately, unless this structure is fundamentally solved, such items are difficult to sustain. In fact, the solutions that have survived are closer to indirect B2B solutions provided in bulk by companies or schools rather than individuals.

Again, What Problem to Solve

At the beginning of class, the professor gave several examples of items worth noting at the time from the perspective we've been discussing—exploring problems in areas where inefficiencies occur structurally and haven't been sufficiently addressed.

• Household accounting or fintech: Inefficiencies remaining throughout the payment and settlement process
• Schedule coordination: Communication costs that grow exponentially as more stakeholders are involved
• Collaboration tools, productivity apps: Especially for mid-sized companies, internal processes and system maintenance often lag behind as they focus on growth
• Energy industry: An industry where government regulations strongly apply, but structural transitions and growth are simultaneously occurring

We decided at the end of a meeting to shift direction and look for industries with structural inefficiencies, targeting B2B. Rather than capturing specific problems or ideas, we decided to first find industry groups. The industry groups we reviewed in that process included:

• Apparel (sewing) industry
• Food service industry
• Automotive parts industry
• IT companies
• Campus education centers
• VC
• Healthcare industry

Among these, the sewing industry was what we first began to look into deeply, and this ultimately became the choice that led to our final prototype. Once the direction was clearly set, the subsequent process proceeded very quickly and systematically. Our project ultimately received both votes from participating students and the professor's selection, and looking back, this wouldn't have been easy if we hadn't chosen the sewing industry. I think this was because we ultimately followed the professor's repeatedly emphasized feedback to "find a problem of a size you can solve" and chose the area our team could handle best. I'll discuss the professor's feedback on 'finding a problem of a size you can solve' in more detail in the subsequent process.

There were three main reasons why our team could handle the sewing industry relatively effectively. First, one team member was a fashion & textile student who was studying related content in a professor’s class focused on smart factories, which address structural problems in the sewing industry. Another team member's parents worked as a manager in the industry, so we could directly hear about problems actually experienced in the field and quickly receive realistic feedback on the solutions we were conceiving. Finally, I had spent the previous vacation interning in a lab conducting computer vision research, having acquired both basic theory and hands-on skills to handle it to some extent. As these experiences combined, the sewing industry became not just an 'interesting problem' but a problem our team could push through to the end. This is why I emphasized earlier that accurately identifying the resources and capabilities the team has is important.

Finding a Problem of a Size You Can Solve

The subsequent process was a repetition of interviews and feedback. In fact, if we've reached this stage, the subsequent progress will be sufficiently possible somehow. So what's more important is what the professor said: finding a problem of a size the current team can solve—that is, how to find a problem that four college students can solve. Here, I want to briefly organize how we narrowed down this problem.

The apparel industry is a very large industry extending from fabric production to cutting and sewing processes, to the final distribution of produced clothing. The problem is where to step within this vast industry. Below is part of the records I organized for team meetings. Through these materials, I could confirm the thought that inefficiencies in sewing factories might be an unsolved problem, and that this problem is deeply connected not just to improving work efficiency but to industrial structural and social problems.

I decided to look more into inefficiencies in sewing factories and checked additional government reports and research papers. Textile and fashion industry reports presented macro alternatives like automation, smart factories, and expansion of foreign workers, and a survey paper on clothing brand ordering methods relatively well organized the inefficiencies and inconveniences between factories and brands. By checking these contents and cross-validating them through interviews with the professor and industry practitioners, we came to focus on defects occurring in the sewing process, particularly in sewing factories, among the long processes of the apparel industry. Whether clothing is well sewn is difficult to check during the sewing stage, and is mainly checked by skilled workers' visual inspection at the QA stage when finished products come out.

In this way, I was able to dig into the apparel industry that I had no knowledge of and discover one small problem.

Once the problem is this specific, we can easily think of clear directions for how to solve it, and likewise can clearly explore past technical attempts or related technologies to solve it.

Now I could see how to address this problem. While changing the entire sewing industry might be impossible, reducing sewing errors in the sewing process on-site is a problem we can sufficiently address. We set our final hypothesis that we could solve the problem by creating a solution that improves this inefficiency and selling it to companies operating sewing factories.

The Story After

The stories that follow are, strictly speaking, somewhat distant from the core this article wants to convey. However, they were a process of verifying whether the problem we defined was actually an implementable problem, so I'll only briefly touch on them.

I believed that we needed to detect the results of the sewing process in real-time to give immediate feedback to workers, while also being able to be introduced into existing factory systems without difficulty. Considering these conditions, I judged that an on-device solution directly attached to sewing machines was more suitable than a centralized structure. To this end, I directly constructed a sewing dataset and, based on that data, trained a YOLOv5 model to verify at a prototype level the technical feasibility of detecting sewing errors in real-time.

Additionally, I selected hardware needed for the prototype such as Raspberry Pi modules, accelerators, and camera models, calculated costs, and hypothetically estimated how many sewing machines the solution would need to be introduced to for the profit structure to work, and at what price point we could sell to companies. As a competitive model, I presented Mitsubishi Electric's sewing machines with sewing error detection functions, emphasizing that our solution could be introduced at relatively lower cost and entry barriers since it could be attached to existing sewing machines without replacing specific equipment.

Conclusion

I'll conclude the article with stories the professor shared at the end of the semester.