I want to become the global No. 1 Answering AI (QA) company in the enterprise market.
Forty to Maru (42 floors) This is the aspiration that Kim Dong-hwan has. Forty Two Maru is a startup founded by search experts. It was established on October 22, 2015. The four co-founders, including CEO Kim, are from Empas, who led the domestic search market 20 years ago. CEO Kim, who was an engineer, thought that deep learning would change the paradigm, and established Forty Two Maru, a deep learning-based search company.
CEO Kim said, It is a company that experts who have more than 20 years of commercialization experience in large-capacity search portals joined forces.
Fortitumaru is a combination of the Arabic numeral 42 and Maru. Maru is a pure Korean word for the top or the top. 42 is a word from British science fiction (The Hitchhiker's Guide to the Milky Way). Like 'Solomon's Ring', it symbolizes absolute wisdom.
The word '42' is the ultimate answer to all the world's questions when the supercomputer (Deep Thought) in this novel returns after traveling through the universe for millions of years (7.5 million years). CEO Kim explained, It is our company's job to find answers to all the world's questions using deep learning.
Forti Two Maru's main business is 'Answering AI'. 'Answering AI' is also called 'QA' because it answers questions immediately by using deep learning with AI-based natural language processing. While the 2nd generation search is finding, the 3rd generation technology, QA, is different from the existing search in that it provides only one answer.
CEO Kim said, In conventional search, you don't find the answer right away when you type in a keyword, but you have to show the search results and click to get the answer you are looking for, said CEO Kim. In this regard, I use the term QA instead of search.
Forty Two Maru, pioneering and leading the answering AI market, entered this field in earnest in 2017. Its technological prowess has already been recognized worldwide. In November 2018, it took first place jointly with Google's AI team in 'SQuAD2.0', a global machine reading comprehension (MRC) competition hosted by Stanford University in the United States. Also, in 2020, it ranked first in 'GLGE', a language understanding contest hosted by Microsoft (MS) of the United States. In both competitions, Forti To Maru is the first time a domestic company has won first place. In addition, it won the AWS AI Challenge Award hosted by Amazon (2018), and ranked first among Korean AI companies in the Asia-Pacific high-growth company selected by the British Financial Times (FT), a world-renowned authority.
Our goal is to become the global No. 1 QA company in the enterprise market.
CEO Kim said, Whether it is a startup or a large company, among domestic SW companies, there is no company that has made a proper name in the global market. He said, We want to try this. Below is a Q&A with CEO Kim.
-I am curious about the background of the founding of Forti Two Maru
After graduating from the master's degree, I joined Empas, a general search portal company at the time, as a military service (special exception to military service). I worked at Empas for about 13 to 14 years. Then, in February 2009, it was integrated with Nate operated by SK. Started as a developer at Empas, SK I also served as the head of the search business division that oversees the entire business. I have had various experiences such as development, strategy, planning, marketing, sales, etc. I was lucky to have done everything that other people had to do in their lifetime. So, after taking a break for a while, IT I was going to do something other than that, but I fell in love with deep learning and started Forty Two Maru.
-Did you become obsessed with deep learning and start a business?
Yes. Artificial intelligence entered a new era with the advent of deep learning. At first, I was not very interested in deep learning. I even thought that it was a scam. But it was not. I thought it could be a technology. It wasn’t the artificial intelligence technology I learned in college. I thought that the technological paradigm was completely changing. While studying deep learning, I thought, ‘This is going to last a lifetime. I thought there must be, so I jumped into the business again. There are 4 co-founders, but all of them still remain. The company that made a seed investment for us at the time was near Seoul National University Station on Subway Line 2, so we started our business here..
-What products and technologies do you have
There are five major ones. AI-based Q&A solution (QA42), text analysis and insight solution (TA42), conversational AI chatbot (CHAT42), leading natural language processing technology (NL42), and search technology (Search 42). (Answering AI) was firmly established as a business item in 2017. Based on the references so far, it is also supplied as a recent package. It continues to advance its functions by finding the customer's pain point. The customer group is also wide. Our first customer was Daewoo Shipbuilding, and we are expanding our clients to include shipping, manufacturing, finance, law, and healthcare.
-What are your main customers?
The first customer, Daewoo Shipbuilding & Marine Engineering, is diverse, including Shinhan Life, Hana Card, Shinhan Bank, Shinhan Card, KT, Law Firm Sejong, Hyundai Kia Motors, LG Electronics, SK Innovation, LG U+, Ministry of Culture, Yongin Severance Hospital, NIA, etc..
-Is there a company like Forti Two Maru in Korea?
Among domestic startups, there is no other place with a broad spectrum like us in the field of AI answering.
-What new business area are you looking at?
Speech recognition (STT) and OCR. I haven't done this in a while. This field is technically leveled up. It is a market where commercialization rather than technology, that is, who makes how well the learning data is more important. Learning data for voice recognition is extensive. The US has also invested 1 trillion won a year in this field and 10 trillion won for 10 years, but it is not enough. We have just started developing the audio field. QA engine is sufficient in these fields because we have the best technology in Korea. I think we have a competitive edge.
-How much investment did you attract
We received seed and Series A. We didn't receive much investment because of the sales. Spring Kemp, a VC specializing in Naver-affiliated startups, KDB, Hana Financial Investment, IBK IBK, Webcash Financial Group, and Hana Bank invested.. There are foreign investors. Techstar, a British global accelerator, invested. We look long and big. We grow fast and do not want to exit quickly.
-I heard that you are looking at Europe rather than the US as an overseas market...
Techstar, one of our investment companies, is the second-largest accelerator in the world after the American Y Combinator. In Europe, Techstar is more recognized than the Y Combinator. Europe is a new market based on deep learning. In the US, Google and Microsoft are competing with B2C rather than B2B like us. Google has accumulated a lot of data. It takes a lot of time and money to convert it all to deep learning.
Since we are a start-up and the market is still in its infancy, we are trying to compete in the enterprise market and in Europe in order to gain an advantage. By the way, the search in Europe is weak. Google has taken over, so European companies don't search. If we make a reference in London, we will make a successful reference in Europe before the US, such as entering Paris. Our strategy is to pick out successful domain flags by industry. Although the United States has only one country, it is more difficult to do business as each industry has different dominance regions, such as Detroit for automobiles and New York for finance.
-It was said that the originally planned European expansion was hindered by the corona virus
I received an investment from Techstar and received mentoring for 5 months in the UK in 2018. At this time, I thought about listing on the NASDAQ. Rather than doing an IPO in Korea, I succeeded in Europe, mainly in the UK, and based on this, I went to North America and challenged the NASDAQ. I set the goal at the time. I also set up a branch in the UK in 2019. However, with the outbreak of Corona, this plan came to a halt. I thought that Corona would go into a long-term battle, so I turned to a domestic IPO rather than the Nasdaq.
-Recently, I participated in Tech Week held in England again. Did the European challenge start again
A few weeks ago, I participated in London Tech Week again after two years. London Tech Week is held for a week in early to mid-June every year. Various events are held, one of which is the 'AI Summit', which is exclusively for AI. Start of London Tech Week Europe's B2B market opens. As you can see from Deep Mind, Europe's AI source technology is fine, but its commercialization is lagging behind Silicon Valley. There were no companies that stood out yet. I saw startups that do the elements of what we do, but I did not see a company that does total solutions like us. It would be good to collaborate with a local company in the UK.
-When will the IPO be held?
After changing the direction to a domestic IPO, we selected a host company at the end of 2020. We are preparing hard for an IPO at the beginning of next year. The overseas market is reopening, and we are still thinking about various options.
-What are the domestic and international market and technology trends in the field of Answering AI
If you look at computer engineering, it started with language processing or text analysis and developed in the order of signal processing, voice, and video. However, deep learning is the other way around. I've studied a lot of visions and images, and this field has been leveled up. On the other hand, language processing The page came later than images and videos, but it is being actively discussed recently. Language should reflect human intellectual ability in addition to what is visible. For example, articles and interviews come out differently depending on how the person is writing. It is more difficult because it is a field where intellectual ability and thinking must go together.
-What is the technological competitiveness of Forti Two Maru?
I won 1st place with Google in November 2018 at an event hosted by Stanford University called SQuAD. Leaderboards change frequently these days. We are the first Korean company to take first place in this field.
-How is the competitiveness of manpower?
Forti To Maru is a startup, but it is a company created by members who have more than 20 years of experience in commercializing large-scale search portals. The managers and team leaders who worked with me at Empas and SK are here. Each member is a team leader at Naver and Kakao. Teamwork is important for startups. Because I worked with me for 10 or 20 years, we get along well. Currently, we have about 90 employees.
-I heard that you also won 1st place in the language AI contest hosted by MS
Last year and this year, we took first place. Microsoft took first place at the Microsoft event for language generation, i.e. language generation model. Language generation is more difficult because you have to create something rather than find it in a document. Language Language generation is a new technology approach in the processing area. It is an area that has not been commercialized in the past 60 years. It is the first time for a Korean company to win first place in the MS competition.
-What is the difference between AI search and conventional search?
If you go AI-based, you don't use the word search often. It's qualitatively different. Existing search doesn't find you right away when you type in a keyword (natural language), but shows the search results and clicks. AI-based QA (Answering AI) Unlike the existing search, when a query comes in, it only finds an answer. It finds it immediately. So instead of searching, we use the words QA or answering. Previously, we used the word search. Because it is information search and information search. The term QA was not widely used in the industry as well as academia in 2017, when it first started. But now QA and MRC have become the standard. Our vision is to become the world's best QA company.
-What are the upcoming products?
We plan to expand or internalize cooperation in voice recognition and OCR, which we have been collaborating with. The market for TA that analyzes text is now in its infancy. The TA we are doing is text QA that analyzes text, such as analyzing call center consultation data. In the future, Visual QA (VQA) that finds an answer if you ask a question with an image or video is likely to appear. We are preparing for this. Also, we are developing a technology that has not yet been commercialized anywhere in the world as a project of the National Intelligence Service (NIA).. We plan to carry out a pilot service by the end of the year and commercialize it next year. I was surprised to find such a task at the NIA.
Company vision in 5 or 10 years
I want to become the global No. 1 QA company in the enterprise (B2B) market. I have no intention of doing B2C yet. Also, among domestic companies, no start-up or large company has done it properly in the global market with software. I want to try it. This is possible because there is a learning mechanism called deep learning.
-Let's ask a private question. If you invite someone to a dinner, who and why?
They are the employees. They are people who have been recognized by large corporations, but they believe in me and are struggling. It would be convenient if our company only had one field (domain), but the resources are insufficient, but the employees are having a hard time because I keep trying to do new things. (Laughs)
-What books and movies have you been influenced by or your role models?
There is no role model. As a film developer, I tend to like sci-fi movies that inspire new ideas. I remember The Hitchhiker and the 1982 sci-fi movie Blade Runner. Sci-fi movies help us imagine the future society as a developer. Recently, I have been reading management-related books. Rather than reading deeply, I tend to quickly read only the parts I need.
-What are your hobbies and mottos?
Before starting the business, I was interested in various sales, but now I have no particular hobbies. Recently, I have been walking a lot. I don't play golf. I used to ride a bike or mountaineering. (Laughs). I think walking is good. I think a lot while walking. If I don't take time like this, I think it's just for living and for work. So I make time for walking on purpose.
In severe cases, I have meetings every 15 minutes. I'm really short on time. I have no particular motto, but I have been interested in Eastern philosophy since childhood. Among them, there is a lot of interest in Lao-tzu's inert nature. Unwitting nature is not doing nothing. It is to live according to the rules of the merchant's medicinal water (上善若水). The company is run according to the rules and performance is achieved. If you have a solid foundation, you will get an opportunity someday and you will be recognized.
Comments
Post a Comment