"Although ChatGPT and Gemini also have translation features, DeepL is specialized in the professional translations that corporations need."
Sebastian Enderlein, Chief Technology Officer (CTO) of DeepL, responded to a question about what sets DeepL apart from its competitors during a press conference held on the 21st at the Grand InterContinental Seoul Parnas in Samseong-dong, Seoul. He said that while the universal translation features of major artificial intelligence (AI) companies, such as OpenAI's ChatGPT, Google's Gemini, and Anthropic's Claude, are like a "Swiss Army knife" (meaning tools that can be used for multiple purposes), DeepL provides advanced customized translations for corporations. Notably, in Korea, it offers specialized solutions for legal document translations to Shin & Kim LLC.
CTO Enderlein noted, "As DeepL is a service specialized in language and translation, we conduct large-scale blind tests before launching specific languages, and we are designing our solutions to eliminate the need for follow-up tasks after translation." He explained that typically, after using an AI-based translator, there is a subsequent process in which a person reviews and edits the translated material, yet DeepL is focusing on minimizing this post-editing.
He said, "People use translators to reduce the time required for translation, so whether follow-up translation or editing is necessary after the translation determines the competitive advantage of the translation solution," and added, "In blind tests, the preference for DeepL's translated outputs was found to be higher compared to those from competitors."
DeepL, an AI company specializing in translation, was established in 2017 by Jaroslaw Kutylowski in Cologne, Germany. DeepL offers various AI-based translation tools, including text and voice translation, AI-based writing solutions, and an advanced application programming interface (API). Its flagship product, the DeepL translator, enables real-time translation in 36 languages. Currently, it has over 200,000 corporations and government agencies as clients across 228 countries worldwide and entered the Korean market in 2021.
CTO Enderlein stated, "Korea is a key market because it is home to multinational manufacturing and distribution corporations that require high-quality translation solutions," and he expressed a desire to help Korean corporations break down language barriers and expand overseas.
DeepL has partnered with Saltlux Innovation to build a multilingual translation AI ecosystem among domestic AI companies. It is currently conducting a joint project with IT service company Ativerse to enhance the global communication capabilities of domestic corporations.
DeepL has upgraded the functionalities of its voice translation solution, "DeepL Voice," which was launched last November. Following the integration with Microsoft Teams, "DeepL Voice for Meetings" has now been incorporated into the video conferencing platform Zoom. From now on, Zoom meeting participants will be able to speak in their native language or their preferred language and see the content displayed as real-time translated subtitles.
CTO Enderlein remarked, "Until recently, English was the common language in the business world, but with the development of language AI, corporations no longer need to speak English to operate on the global stage," adding, "Everyone will be able to speak and write in the language they can use most naturally and collaborate seamlessly."
"The languages supported by 'DeepL Voice' have expanded from the existing 13 to a total of 16, now including Chinese, Ukrainian, and Romanian. Subtitle translation has also increased to a total of 35 languages, including Vietnamese and Hebrew, which were recently added to the DeepL platform."
CTO Enderlein identified real-time translation as another strength unique to DeepL. He noted, "Last month, we introduced NVIDIA's DGX SuperPOD supercomputing infrastructure, becoming the first European company to do so, which boosted web translation speed by 10 times." He explained that with the introduction of the NVIDIA SuperPOD, the time required for full web translation was reduced from 194 days to 18.5 days, and the total volume of text processed increased by 30 times compared to before. He also mentioned that for Marcel Proust's "In Search of Lost Time," which is considered the longest novel in the world, the translation time dropped from the original 0.95 seconds to 0.09 seconds.
He emphasized, "We will utilize next-generation computing infrastructure to provide more personalized translation solutions and take the lead in transforming the $60 billion language services market."