Startup Log: Bayesians

On April 17, 2025, our team incorporated in the United States. Although the final approval came the next day, we designated my birthday, April 17th, as our Day One. This post is a record of the 90-day journey of our team’s formation, from January 17th to April 17th, 2025.

The Beginning of a Legend

In mid-January, after parting ways with the connection I made at Antler, I became an individual again. I continued to work alone afterwards, but looking back, I think I felt a sense of hopelessness at that time. I didn’t think it was absolutely necessary to work with someone, but working alone, I felt like I was only thinking within the limits of my own abilities. The business I was considering at the time was a personal information management service. Among them, I was thinking of summarizing and organizing newsletters, because that was a problem I had. When I felt that my problems were limited to what I intuitively felt and understood, I started to think, ‘a co-founder is essential for me.’

I immediately posted on LinkedIn asking for coffee chats. I wasn’t thinking of finding a co-founder right away. As I wrote in the post, it was a time when I thought that the process of starting a company should be viewed with a very, very long-term perspective, so I wrote the post with the thought of charging up my ideas and making as many connections as possible.

Amazingly, it got 13,790 impressions. I think it was the most widespread post I’ve ever written. As much as it spread, a lot of people asked for a coffee chat. In one week, I made 22 appointments with 19 people.

There were also requests for coffee chats like ‘looking for a free developer.’ “I don’t have a developer,” “I have something I really want to try.” In such cases, I told them I wasn’t joining the team and asked if a coffee chat would still be okay. Without exception, every single one decided not to meet and wished me well.

MJ was one of the 19 people, and someone I met twice. MJ was fashionably late for our first meeting and bought lunch. When I met him then, we talked casually, and as always, I tried hard to see his motivation for starting a company and the size of his dream. For our second meeting at a cafe, we each prepared answers to First Round’s 50 Questions for Co-Founders and compared them. Looking at my personal retrospective from the 4th week of January, I felt that our answers to the 50 Questions matched well, and I was satisfied with our expectations for each other and our attitude towards entrepreneurship. But the most satisfying point was that he wanted to put a 10-15 year cliff on the co-founder. It seemed like he was just as interested as I was in preventing the co-founder from leaving, and I thought, ‘this person will live and breathe entrepreneurship until the end!’ We started working together in earnest the following week.

Agreed-Upon Goals and Team Color

We both had the basics of going to the US and building a global service from Day 1. Our criteria, since we hadn’t decided on a specific domain for this business, was to solve a problem that could become at least a $1B company. We figured that if the global market is around $50T to $100T, we could reach that goal regardless of the segment we entered.

The first area we tried to work on together was the researcher’s domain. We saw that, just as Cursor did, we could create an optimal UX in the pattern of how researchers read and organize papers. After interviewing about 10 researchers, we confirmed the same pattern and approached the first product as a kind of Notion for Researchers. We built and deployed a service with an editing tool for organizing like Notion, a PDF viewer, and a RAG that understood PDFs very well in just 3 days. We talked to some of the people we had interviewed again, but the value we were delivering was very vague. Looking back now, I think we were also wondering, ‘is this really the right product to optimize the pattern we saw in the interviews?’

When I first thought about this area, I wondered, ‘is it right to dig into such a narrow market…’ And I still have that thought to some extent. As a result, in the process of getting more specific in this area, our customer segment became even narrower, and we judged that the scale would not match our goals, so we moved on to another item.

What was meaningful in this process was that I got a rough idea of what it would be like to work with MJ.

• We are very impatient people. The team has a driving force to make everything other than the conclusion secondary and rush towards the conclusion. This quality doesn’t always yield positive results, but I think it will be quite useful in the future.
• We both use reason well. We know how to think about what kind of evidence is needed to be rational when defining a problem or persuading others.
• We seem to like working to a similar extent. When meeting co-founders, everyone always says they like to work, but when you look at how much time they need to work to maintain a sustainable level, it’s rare to find a perfect match. I wonder if it’s not sustainable to work more if you really enjoy it, and in that sense, we seem to enjoy working similarly.

An Area We’re Okay With for a Long Time

After that, we thought up and validated ideas very quickly. We had about five ideas before the one we’re working on now, and we checked one item almost every week. I was continuing with the thought that finding an item wouldn’t be easy, but it was a shame that when the item changed at the domain level, the team’s know-how and lessons learned, even if small, became meaningless. So I thought it would be better if we could pivot and move forward within an area that we were okay with for a quite a long time.

The first domain we worked on for more than 2 weeks was ‘animation.’ 2D Animation is a content area that MJ is attached to, and he had conducted related AI research before graduating. I, too, haven’t reached the level of an otaku, but I’ve enjoyed watching 2D animation since I was young. At first, we thought about an AI tool for animation production, but by combining the thoughts within the team and the thoughts of those who were helping us, we are moving in a direction that can also reach the consumer side.

The consumer side is more an area of fun than a problem area, so I think it requires faith rather than logic. There is a belief that if we can reduce the cost of creating good content by 1/10000 based on the founders’ personal experiences, we can succeed, and there is a belief that we are the people who are good at this. Right now, we are adopting and automating Wattpad’s multi-million view works with AI Animation.

We are collaborating closely with the original author, and fortunately, they are reacting positively to the initial quality of the AI Anime.

I feel more and more that we are approaching the ‘on-time’ for a major service to emerge in the AI video space. Even if the quality we produce right now is lacking or the cost is high, like GPT, it will soon drop to the consumer level, get faster, and get better. I feel like we are preparing for that on-time.

Succeeding or in the Process of Succeeding

I really like the way Noh Hong-chul thinks. On Pani Bottle’s channel, what Hong-chul said while traveling with Bottle almost perfectly matches my thoughts. To the question, ‘How do you only do what you want to do and everything works out?’, he replied, ‘I did it until it worked out.’ Until you give up, you are in the process of succeeding. In that sense, a team that doesn’t seem like it will give up is the same as a team that doesn’t seem like it will fail. In my opinion, our current team is a team that is unlikely to fail unless a situation arises where it is impossible to work due to serious injury or illness of an individual or family member. As we move towards success, we will be joined by many people, and I hope they also get the feeling that we are a ‘team that will make it happen no matter what.’

Day 1

For a combination of reasons, we quickly set up a US corporation. Launching the corporation on my birthday seems to create another attachment. Bayesians is a name I had been using for organizations in various situations since starting my company last year. It’s closely related to AI, but I chose the name because I believe that people who can think about conditional probability can increase their win rate in decision-making. MJ said the name sounded good, so we’ve been using it as our team name regardless of the item. We’ve been in stealth mode until now, but I think we’re now in a state where we can definitively reveal the team.