[Founder Interview #4] Tsuyoshi Tateishi (SpeakBUDDY) – Part 1
From 280 Points on TOEIC to a Perfect Score and 5,000 Hours of Studying, Forming a Belief in the Necessity of AI English Conversation
In December 2025, D-POPS GROUP invested in SpeakBUDDY Ltd., the developer and operator of the “SpeakBuddy” AI English conversation app. (Read more about it here.)
How did someone who was once considered the worst English student in his grade go on to build a cutting-edge, AI-powered English conversation service? In this three-part series, we look into the entrepreneurial journey of SpeakBUDDY’s founder Tsuyoshi Tateishi and the future he envisions for language learning.
In Part 1, we explore his incredible transition from a low TOEIC score of 280 to 5,000 hours of dedicated study, his experiences at a foreign investment bank, and how a trip around the world sparked the idea for AI-driven English conversation. (This interview was conducted in March 2026.)
◆“Worst at English in the Entire Grade”
Sugihara:
To start off, could you tell us what gave you the initial push to develop the SpeakBuddy app?
Tateishi:
First of all, the biggest reason I chose the field of English was because I was actually terrible at it during my student years. English requires consistent, cumulative effort, but I hit a wall in my first year of junior high school and just couldn't keep up with my classes after that. By high school, my teacher told me I was "the worst at English in the entire grade."
Sugihara:
So the student who was the worst in his grade became the CEO of an English conversation app company, ha ha!
Tateishi:
Exactly. That was my reality in high school, but the turning point came during my job hunt in university. I received a tentative offer from a foreign investment bank, and starting my career there was what finally forced me to start learning.
Sugihara:
It was a foreign-owned firm, so didn't they interview you in English?
Tateishi:
Oh, they did. The final interview was in a boardroom with the then-CEO of Citigroup Global Markets Japan and a row of executives from various departments. At the time, they were testing whether candidates could handle that kind of pressure. Since investment banking is a business where people are the only real assets, they were looking for character.
That 10-on-1 interview was mostly in Japanese, but at the very end, the HR Director said, “Alright, I’m going to ask you a question in English, so please answer in English.” Up until that point, all the interviewers were Japanese, so they were probably assuming, “He graduated from Keio University, so he can probably handle it.”
But in fact, my English was non-existent. My TOEIC score was only 280, so I knew I wouldn’t be able to understand any questions if they asked me in English, so I had prepared exactly one response in my head: “I can’t speak English. But I will study hard. So, no problem!”
I said it with total confidence, but in fact, that single sentence was the only English I had learned. Back at that time, just memorizing that one phrase had been a struggle for me! The entire room of interviewers froze. Everyone was like, “Wait, what? You applied here with that level of English? This is a foreign firm…what are you expecting to do?”
However, this was during my third year of university, so I told them, “I still have one year until I graduate.” At the time, I had just become the youngest person that year to pass the CPA exam at age 20. I was studying 14 hours a day…other than study and sleep, I would spend only two hours a day to shower, eat, and commute. I was used to finishing an entire textbook in a single day. So, I truly believed it when I told them, “I have a year, and that’s enough time for me, seriously.”
At the time, I figured if I could get a 750 on the TOEIC, I would be basically fluent. Compared to the CPA exam, where if you fail, you have to wait another year, I felt no pressure at all. When I said, “I’ll start tomorrow”, they looked at me and said, “You do look like the type of guy who would actually do that.” Still, I walked out thinking I had definitely failed the interview, since it was a global firm and I couldn’t speak a word of English.
But later that day, someone from their HR called me. Apparently, the CEO had been my biggest advocate. He told them, “Hire that guy.” The HR worker asked me, “What on earth did you say to the CEO?” Now, what I had told the CEO was this: “I might not speak English now, but through my CPA studies, I realized there isn’t that much difference in the capacity of human brains. It’s all about having grit. I know for sure that if I put my mind to something, there’s nothing I can’t achieve, and I will absolutely get this done.” The CEO decided, “Let’s hire one interesting guy like this”, and that’s how I got the job.
◆The Problem for Japanese Who Study English for 5,000 Hours
Tateishi:
Over the following years, I reached a perfect score on the TOEIC, passed the Eiken Grade 1 (the highest level of Japan’s most popular English proficiency test), and logged a cumulative total of approximately 5,000 hours of English study.
Through those 5,000 hours, I had a painful realization: learning English would be far more difficult than I thought it would be when I was job hunting. I realized that this might be the single greatest challenge facing the Japanese people. I struggled with it while at that foreign-owned firm, and even after I moved to a Japanese brokerage, I was stationed in Hong Kong where I had to use both English and Chinese. If you can't use the language, you simply cannot do the job.
I learned enough Chinese in Hong Kong to handle daily conversation, but if people ultimately can't understand what you're saying, you can't conduct business. Japanese people are excellent workers, but just because their English isn't great, my compatriots were being looked down on by people from headquarters or other overseas branches, and I couldn't stand it. During global training sessions and such, Japanese participants may not be able to speak English at all, but when it comes to the actual work output, they are consistently the best.
Changing the reality that Japanese people are brilliant but are dismissed because they can’t communicate in English became my core motivation. I spent 5,000 hours to get there, but that’s not a realistic path for everyone. I want to use the power of technology to drastically shorten that time. This is the fundamental reason I chose the field of English education and began developing SpeakBuddy.
◆Making a “Doraemon” Who Can Speak English at Home
Sugihara:
Did you always have the desire to be an entrepreneur, even as a student?
Tateishi:
I did. I remember asking during my job interviews, “If I work at this investment bank, will it help me when I start my own company later?” But after seven years of intense, high-pressure work, I had completely forgotten that dream. However, as my desire to create a service that truly helps society grew stronger, I remembered, “Oh right, I actually wanted to start a company!” So, I quit as soon as I returned from my overseas assignment.
I decided to quit first, and think of a business plan later. So, with my back against the wall, I finally went on the world tour I had always wanted to do, and I mapped out my business plan while traveling. While hopping from country to country, I was also teaching myself app development. The English learning app I built during that trip actually reached number 1 in the overall App Store rankings.
Sugihara:
Hitting an overall ranking of 1 is incredible! Was that app the prototype for what SpeakBuddy is today?
Tateishi:
Yes, I’d say that was the prototype. During my world tour, I also did a short-term study abroad program. Even when you are overseas, you find yourself wanting someone to practice English conversation with. But making friends in English out of thin air is incredibly difficult. People always say, “Just get a foreign boyfriend or girlfriend”, and I actually tried to do that while I was abroad, but it’s not as easy as it sounds, ha ha!
While I was abroad, I thought, “If only I had a Doraemon* at home who could speak English with me, that would be enough.” That desire to create a Doraemon who speaks English was the starting point of our service.
*Note: ‘Doraemon’ is a beloved character from a Japanese children’s cartoon who always helps the other characters with their various problems.
I included a voice recognition feature in that first app, and I’m proud to say it was likely the first app in Japan to feature English voice recognition. However, back in 2014, the accuracy was still quite low. But over the next year or two, I felt the technology improving at an exponential rate. I became convinced that in the future, I would be able to create a real AI English conversation partner. In 2016, I stopped everything else I was working on and focused all my resources on AI English conversation.
◆The Problem with AI-related Branding in 2016
Sugihara:
When you released the precursor to your current app back in 2016, were you already using the word ‘AI’?
Tateishi:
Yes. Even at the 2016 launch, we were calling it ‘AI English Conversation’. Since we had no capital at the start, we began with a crowdfunding campaign. It went surprisingly well—we raised about 4 million yen—and we were already positioning it as an AI-driven service even then.
Sugihara:
The speed of AI adoption and evolution lately is staggering. While language models have developed explosively in the last year or two, you were really at the forefront.
Tateishi:
We were the only ones in that space back then. But honestly, the immediate aftermath of the release was a struggle. People in 2016 were very vocal with their skepticism, asking, “Where exactly is the AI in this?” They’d complain, “I’m talking, but this doesn’t recognize my voice at all!”
Back then, AI conversation meant little more than speaking to a character with an electronic voice. The conversational AI was still in its infancy, with a simple tree structure where we programmed specific responses to specific phrases. You couldn’t engage with it via truly free conversation like we have now. In the end, because we were getting so many reviews asking “Where’s the AI?”, we actually removed the word ‘AI’ from our branding around 2017. Having AI in the title seemed to raise expectations so high that it led to people leaving 1-star reviews, so we decided to take it out for a while.
Then, around 2019, speech recognition technology started to improve dramatically. We also built our own proprietary speech recognition engine, which boosted our accuracy significantly. From that point on, the quality of our conversational AI was able to truly shine.
Sugihara:
When I try using features like BuddyChat now, the AI responds naturally even if I say something completely random or trivial. It clearly isn’t just simple pattern recognition anymore.
Tateishi:
Yes, it can handle tangents and keep the conversation going. I feel like I can finally say “This is AI” with my head held high. Back in 2015 and 2016, we were right in that peak of inflated expectations on the hype cycle, which was followed by a valley of disillusionment. During that valley, people kept telling me, “AI is a joke; it’ll never work.” But gradually, it became the real deal.
Sugihara:
For someone who has been working on this for a decade, do you feel like it's about time for the AI boom to finally sweep across the world?
Tateishi:
In a way, yes. It feels like things are moving exactly as I expected, though there’s a bit of a time lag in terms of when the world finally noticed. To be honest, even with GPT, we were looking at GPT-2 and saying, “This is going to be incredibly useful.” So by the time GPT-4 arrived, our reaction was like, “Well, it’s already been impressive since the beginning.”
It’s only been in the last few years that speech recognition, Natural Language Understanding (NLU), and Personalized Learning Logic have finally synced up. Because the level of inference has jumped so significantly recently, we can now provide high-level free talk like BuddyChat. It feels like the times have finally caught up with SpeakBUDDY.
~To be continued in Part 2~
Interview conducted by D-POPS GROUP's advisor Genta Sugihara.
SpeakBUDDY, Ltd.
President and CEO: Tsuyoshi Tateishi
Address: 3F +SHIFT Nihonbashi Sakuradori, 3-14-3 Nihonbashi, Chuo-ku, Tokyo
Established: May 2013
Website: https://www.speakbuddy.com/en
Next, in the second part of the interview, we discuss:
・The four pillars of SpeakBUDDY's competitive advantage
・Where SpeakBUDDY differs from Duolingo and Speak
・Rapid growth in the corporate market and feedback from business clients
・And other topics
Be sure to check it out here:
https://d-pops-group.co.jp/en/column/speakbuddy-part2/