Transcriptly

How I built $2B AI Translation Startup | DeepL, Jarek Kutylowski

EO13:50Added Aug 31, 2026

Today's story is about Jarek Kutylowski, the CEO of DeepL. DeepL is a platform that provides high-quality machine translation and AI-based writing tools, supporting multilingual communication for businesses. It was also one of the few companies to develop a product based on AI long before AI became widely popular. Recently, it concluded a fundraising round with a valuation of $2 billion. Jarek, the CEO, moved from Poland to Germany during his chi

Watch on YouTube →
Contributed by 刘嘉琪

Transcript

Transcript format

  1. Intro

  2. 00:00We've been the first company to the market with an AI based
  3. 00:04neural network translation solution, which just blew away everybody else.
  4. 00:08And we gathered the early adopters that they started spreading the word
  5. 00:12of mouth and it was all about speed.
  6. 00:14If we have come up like half a year later, a year later,
  7. 00:17I don't know if that would have worked so well.
  8. 00:19You have to move fast.
  9. 00:20You have to figure out what is the next challenge,
  10. 00:23approach it, solve it, and then move on.
  11. 00:25Hi, I'm Jared Kotlowski, founder and CEO of DeepL, a company that builds AI that helps
  12. 00:31breaking down language barriers by making translations available to everyone.
  13. 00:34Deepl has grown out of this huge free service that everybody out there
  14. 00:38in the world can use, and this service is being used by hundreds
  15. 00:41of millions of people every month.
  16. 00:43And out of that, we've built a base of over 100,000 companies
  17. 00:47that are working people on top of that.
  18. 00:49A few months ago, we just completed a fundraise which was valued at $2 billion.

  19. Sponsor

  20. 00:56Are you a tech startup founder building something cool?
  21. 00:58If so, listen up.
  22. 01:00EO has teamed up with .Tech Domains to give founders like you
  23. 01:03a rare opportunity to pitch live directly to top VCs in Silicon Valley
  24. 01:07and get featured on this channel.
  25. 01:09Viewed by more than 350K subscribers worldwide.
  26. 01:13The only condition your startup must use a dot tech domain or plan to.
  27. 01:17Upon getting selected to pitch and erase less than $3 million in funding.
  28. 01:21After all, if you're building in tech, dot tech is the most
  29. 01:25relevant domain extension to get.
  30. 01:26It tells the world your website is all about technology,
  31. 01:29and hey, it's cooler than a.com.
  32. 01:31Some of the coolest tech startups are already using dot tech domains,
  33. 01:35and we've been highlighting their stories in Io's original interview series, Beyond Tech
  34. 01:40Now it's your turn.
  35. 01:41Head over to Pitch Tech or check the video description to apply.

  36. Chapter 1. An immigrant who couldn't speak German

  37. 01:51I grew up in Poland, which was at this point in time kind of a country
  38. 01:55in a switch between the communist system towards a post-communist one.
  39. 01:59I got access to technology a little bit later than I think people in the rest
  40. 02:04of the world, and I was just amazed by how much you can achieve with tech,
  41. 02:08especially with software.
  42. 02:09I moved to Germany with my family, have been thrown into school
  43. 02:15very, very quickly.
  44. 02:16I didn't speak basically a word in German, which made me struggle quite a bit.
  45. 02:21I still remember the first day in in class when I walked in
  46. 02:24and I couldn't even really spell my name.
  47. 02:27I had to learn how to survive in an environment
  48. 02:30maybe a little bit more complicated.
  49. 02:32I was lucky to learn German very, very quickly, so that made me feel welcome
  50. 02:37and belonging in that community quickly.
  51. 02:39I learned language and communication is super important in this world.
  52. 02:43If you want to belong to a community, you need to be able to understand each other.
  53. 02:49Basically, from the very early days when I was fascinated by technology,
  54. 02:53this was really the path also that I have chosen in school and later in in studies,
  55. 02:58I majored in computer science.
  56. 02:59I went on to do a PhD in like really theoretical computer science.
  57. 03:04I really enjoyed a very solid theoretical foundation.
  58. 03:07I think doing a PhD really also builds up a lot of in that process, because usually
  59. 03:14you're really thrown into a field of research that is not yet discovered.
  60. 03:18You have to really uncover something.
  61. 03:21You have to go through that process pretty much on your own, at least for myself,
  62. 03:25but also for all of the peers that I've seen built up a lot of resilience for
  63. 03:29for the future of their of their life.
  64. 03:32After my PhD, after my really academic years, I spent a little bit of time
  65. 03:36working for a larger corporation that didn't really fully suit me.
  66. 03:40And then at some point in time, I really, I really realized
  67. 03:42I want to build something,
  68. 03:43I want to be part of something bigger and like really starting a company,
  69. 03:47funding a company, kind of embarking on DeepL journey.
  70. 03:50That was something that that really fascinated me.

  71. Chapter 2. Bridging Language Barriers with AI

  72. 03:56I think in 2016 to 2017.
  73. 03:59There was this great moment when it has become pretty clear, at least
  74. 04:03in the in the academic environment,
  75. 04:06there is a lot that can be done with neural networks in AI.
  76. 04:09So that was an excellent point in time in which we started at the beginning,
  77. 04:13really to play around with the technology and see what we can do with that,
  78. 04:17how that can be applied for some problems that we might be thinking of.
  79. 04:21And I had this background in language. I've been living in two countries.
  80. 04:25I knew what it means to speak different languages.
  81. 04:27I knew how big of a problem that is from a European perspective.
  82. 04:31Each and every time you want to travel to another country, but most importantly
  83. 04:35do business with another country, there is going to be a language barrier.
  84. 04:38If we look at Germany, the companies in the country are selling
  85. 04:42to French customers, they're selling to Italian customers, they're going
  86. 04:45to be selling to Polish customers.
  87. 04:47And you can try doing that just by speaking English all of the time.
  88. 04:53But at the end, every person wants to be addressed in some ways in their local
  89. 04:58language, they will understand much better what you're offering them.
  90. 05:01I think for companies, it's really hard to establish those new markets.
  91. 05:05What you have to do is you have to hire people really in the specific region,
  92. 05:10or you have to find people who are qualified to speak
  93. 05:12in a particular language in your country.
  94. 05:15And that can mean even like doubling your sales headcount, or that may mean like
  95. 05:20adding a lot of customer service jobs,
  96. 05:22for example, for within within your company translation language industry
  97. 05:27is being told to be like 60 billion.
  98. 05:29If that is more efficient, if that is more productive,
  99. 05:31they're going to build better products and be more successful in that market.
  100. 05:35I think the very early days were pretty specific for DPL because, like,
  101. 05:39we knew that this problem of translation that this is that this is a big one.
  102. 05:43I think what we didn't know is whether the technology that we're going
  103. 05:47to be able to build is going to be enough to solve those problems
  104. 05:50and be better than our competition.
  105. 05:52I think that was the unknown, but it was pretty clear that there's
  106. 05:55this big problem that can be solved.
  107. 05:57I think what we didn't know really particularly well also was how to embed
  108. 06:01maybe that technological solution into real life applications,
  109. 06:05how we can go fully to the market.
  110. 06:07And for that, we've just kind of tried to go the path of least resistance.
  111. 06:12We built the technology, we put out a free service that was that was super basic,
  112. 06:17that just gave a very bare bones access to the to the technology itself.
  113. 06:22But at the at the same time was also simple to start using
  114. 06:26as few barriers as possible, like no login, nothing like that,
  115. 06:30just go there and start using the product.
  116. 06:32And for us, that was a great way to validate whether this technology and this
  117. 06:36early product idea actually make sense.
  118. 06:38And if there's market opportunity for that.
  119. 06:41And out of that, we've seen like a very clear signals
  120. 06:44that this is actually what people want.
  121. 06:46And this is actually what, what users what users need through just
  122. 06:50purely looking at the usage numbers.
  123. 06:52That was super simple.
  124. 06:53I think the next step and the challenge there was to look whether this is
  125. 06:57something that people are going to also be willing to pay for, whether there
  126. 07:00is a monetization pattern for that.
  127. 07:03And this is something that we then started doing in 2018,
  128. 07:07introducing new functionality into the product, potentially putting
  129. 07:10those behind paywalls and seeing whether we can convert customers.
  130. 07:15We can convert free users into into being customers.
  131. 07:19A lot of that at the very beginning was really based on a gut feeling.
  132. 07:22And I think at the very beginning you have to have those hypotheses
  133. 07:25which come out of the founding team.
  134. 07:28But I think specifically when you're working in such a slightly
  135. 07:31more consumer ish market at the beginning, you have to rely a lot on quantitative
  136. 07:36data rather than on qualitative findings.
  137. 07:38The more customer focused that came slightly later
  138. 07:41when we started shifting the product towards a B2B and enterprise persona.
  139. 07:45As a buyer,
  140. 07:47I think the biggest challenge as a first time CEO is really making sure
  141. 07:51that you're making your decisions and that
  142. 07:54you're kind of pushing the company at the speed that you could because you don't
  143. 07:58know what the next step potentially is.
  144. 08:00You have to rely on a lot of advice on how the company is going
  145. 08:03to look in the future.
  146. 08:04You have to find out things on your own, like you kind of understand
  147. 08:07the point which you are in and extrapolate to the next point.
  148. 08:12Doing that is incredibly slow. Maybe sometimes.
  149. 08:16If I were now to found another company and do this the second time, I think I
  150. 08:21could be just much, much faster in that.
  151. 08:24Other than that, I do not think that we've as a company
  152. 08:26made too many like big really mistakes.
  153. 08:29I think speeding it all up would just make such a big difference I guess.

  154. Chapter 3. Speed Matters

  155. 08:37If you're in a startup, and especially if you're in a competitive field like ours,
  156. 08:42like with all of the big tech also having their solutions, you always have to grow
  157. 08:46and you have to think about growing fast.
  158. 08:48That is essential for a company for for like even the company's motivational
  159. 08:52health in a way like it always needs to very fast grow all of the time.
  160. 08:57And that made us obviously also scale the business in terms of employees,
  161. 09:01in terms of the number of customers that we have, in terms of the amount
  162. 09:05of products that we are offering.
  163. 09:06And all of that was really tailored to that.
  164. 09:09In 2018, the business was operating profitably already.
  165. 09:12That is just a function of how cost effective a PLG growth motion is,
  166. 09:16where you don't have to hire a lot of salespeople.
  167. 09:19Your customers pretty much come to yourself because they're convinced
  168. 09:22of the product, but at the same time being financially responsible.
  169. 09:26We've been lucky in that way as a company, really, and I think the biggest lesson is,
  170. 09:30once again, it's all about the speed.
  171. 09:32We've been the first company to the market with, with like an AI based
  172. 09:36neural network translation solution, which just blew away everybody else and,
  173. 09:41and made sure that we that we gathered our first user base that that we gathered
  174. 09:46the early adopters, that they started spreading the word of mouth in the world.
  175. 09:51And that was all about speed.
  176. 09:52If we have come up like half a year later, a year later, I don't know
  177. 09:55if that would have worked so well.
  178. 09:57So the biggest lesson is really it's all about speed.
  179. 10:00You have to move fast.
  180. 10:01You have to figure out what is the next challenge,
  181. 10:04approach it, solve it, and then move on.
  182. 10:07If you decided to grow your company very fast, you need to be aware of the change
  183. 10:12that is happening there.
  184. 10:13You have to make sure that you understand what is happening,
  185. 10:16and that you help all of the people in the company be on the change journey.
  186. 10:20I think the biggest problem with moving fast in a high growth company is really
  187. 10:25the fact that everybody in that company has to go through a lot of change,
  188. 10:29because nothing is going to be the same this year as it was the last year.
  189. 10:32Change is hard for us.
  190. 10:34This is something our brains, they do not really like going through, but also trying
  191. 10:39to make sure that everybody knows why we have to go through these change processes,
  192. 10:44why is it important for the company and everybody is on board with that?
  193. 10:48And that is incredibly important if you if you're building an organization.
  194. 10:51But also if you want to go through the change on your own.
  195. 10:54And and I think context and understanding of the why helps
  196. 10:59in addition to that like really a lot.

  197. Chapter 4. Choosing the Right AI: General vs. Specialized

  198. 11:05So there's pretty much two types of AI models that are on the markets right now.
  199. 11:09The very general generative AI models that can do pretty much anything,
  200. 11:13and the specialized models which really focus on creating one particular solution
  201. 11:18at the best quality possible.
  202. 11:20I think the very big generalized models, or those models that you can
  203. 11:23actually use for pretty much anything.
  204. 11:25They have surprised us with their ability to do a wide variety of of tasks really
  205. 11:31I think what needs to be understood that some of those models actually do not
  206. 11:35perform that well on particular instances,
  207. 11:38on particular use cases of problems that we might have, especially in business.
  208. 11:42In the case of translation, for for stable quality and accuracy, that
  209. 11:46is not only on one email, on two emails,
  210. 11:48but like really across the board on a wide range of inputs, and I think this is
  211. 11:53where specialized models can really shine.
  212. 11:55They are usually quality tested for a very particular reason, for a
  213. 11:59very particular use case, and can deliver that quality very, very consistently,
  214. 12:04while at the same time being potentially also more cost effective and quicker early
  215. 12:10to run, which which matters very much in some of the use cases.
  216. 12:13So so I think it's, it's for, for businesses.
  217. 12:15It's always good to take a look at what business problem
  218. 12:18are we trying to solve here.
  219. 12:20What are the general solutions for that versus what are the specialized solutions
  220. 12:25and what kinds of advantages they bring.
  221. 12:28And especially I think the world hasn't changed too much
  222. 12:31in a way that technology itself doesn't yet fully solve the problem.
  223. 12:35You also have to have the product and the integration and the user experience
  224. 12:39and the UI solve for having that problem really meaningfully impact
  225. 12:44your workforce's productivity, efficiency.
  226. 12:47That case, those specialized solutions will usually come with a full suite
  227. 12:52that helps solve that holistically.
  228. 12:54I think if you're thinking about bringing AI into your business, you really
  229. 12:57have to start with the basics like what kind of problem do you want to solve?
  230. 13:02Like what is maybe at the core of the performance of your company?
  231. 13:05What is important for your company and what is potentially working slower
  232. 13:09where you have a problem?
  233. 13:11And starting out with that business problem,
  234. 13:13you can try to to kind of find out what are the potential AI solutions for that.
  235. 13:18I'm not an advocate of trying to to find applications for AI just as a technology.
  236. 13:24I'm an advocate of starting with the problem and then looking for the AI
  237. 13:27solution that can potentially solve that problem in a very good way,
  238. 13:31because through this, you will be optimizing what is really worth optimizing
  239. 13:35in your business rather than trying to apply AI to pretty much everything.