Skuuper brings a DIY spin on Google Translate to the office for law firms, doctors and startups


Professional translation can be a costly endeavor for SMEs or tech startups, even if they are turning a dime or are well-funded. While general services like Google Translate are good for short DIY translations, it is not up to task for long technical jobs needed by law firms and medical professionals. One startup out of Tallinn, Estonia wants to leverage the linguistic skills of the staff you already have to expedite that process with a strategy of so-called DIY translations.

Skuuper‘s Computer Aided Translation (CAT) tech is based on a publicly accessible translation memory (TM) that takes in information based on translations a client company has already done. Any given law firm or investment house might be translating documents on the same topics constantly. Extracting data from those translations and the original versions, Skuuper’s AI can make fast inferences to translate common concepts, terms and phrases that would improve long-tail translation projects.

“Every previously translated text makes a contribution for the next one,” founder and CEO Raul Malmstein told Geektime. “It saves time for the staff or helps to negotiate price with a translation agency. Their multilingual content could be used as an asset for maintaining their documentation flow.”

The Skuuper translation bank is currently available for bilingual translations between six different languages with Portuguese and Mandarin on the way
The Skuuper translation bank is currently available for bilingual translations between six different languages with Portuguese and Mandarin on the way. Image credit: Skuuper

The translation bank (TB) becomes an asset for freelancers and in-house translators or other employees who can speed up the translation process based on how your company has chosen to translate documents before. Their TB already boasts 45 million sentences and 3 million phrases. They are hoping to grow a network of thousands of freelancers whose increased output would contribute mightily to that wealth of data.

The startup currently supports six languages and all possible 15 language pairs among them: Estonian, English, French, German, Spanish and Russian. Portuguese and Mandarin Chinese are also on the way, according to Malmstein. Skuuper’s CAT lets you upload different files (HTML, PDF, .doc and .txt) while maintaining the original formatting. They are building up their repertoire in verticals like law, finance, government and media.

“Technologically we aim to make a language-specific system that distinguishes between them and applies rules accordingly, creating a better user experience. That is why we do not claim you can work with all the languages in the world. We claim you are going to have a quality tool . . . which requires deep [and] knowledge skill to implement,” Malmstein noted.

Vertical-focused translations for clients

Unlike Google Translate or Microsoft Translator, which is more of a machine translation that isn’t at all suitable for longer projects, Skuuper is focusing on concepts that would be specific to advanced topics in verticals like law or finance. Media, business subjects and IT are also big verticals they cover.

He explained, “We actually do not offer machine translation, but rather professionally translated phrases and sentences for several business spheres: law, IT, finance, media, etc, so that people at translation work can avoid retyping of clichés and phrases widely used in documents.”

They aren’t alone in the translation startups space. Portuguese startup Unbabel is making waves with its own machine learning translations.

So far, Skuuper has only raised €125,000, but as a young startup that doesn’t sound like a point against them. They also have offices in Silicon Valley and Russia. Their home base of Tallinn is a thriving center with strong angel networks, development in machine learning and more specifically a growing hub for language tech. Tallinn-London company Lingvist is looking to make a putsch against DuoLingo and Memrise, but is starting first with French and English. We asked Malmstein if he thought there was a particular penchant for language startups coming out of the Baltic state.

The Skuuper translation engine has a massive repositoty of 45 million sentences and 3 million idiomatic phrases already

“As a multilingual society, Estonia definitely is in the heart of localization market, being a member of the EU is even more so — the demand for quick and effective communication grows and fosters a variety of solutions to appear.” Driven by angels who grow their wings at Skype and now with powerful companies like TransferWise and Pipedrive, the Estonian startup ecosystem seems ready to join Israel’s and Singapore’s on the shortlist of small tech scenes whose innovation punches above their respective weights.

“Tallinn is truly an innovative community that offers a healthy environment for hi-tech business. Strong mentorship support, vibrant events for startups and investors, startup-friendly businesses with cooperative policies and of course e-government make Tallinn one of the most attractive places for building a technology,” he stated.

Skuuper is run by Founder and CEO Raul Malmstein, CTO Oleg Urzhumtsev and Linguistic Advisor Liuba Kuncevich. They maintain offices in San Jose, Moscow, St. Petersburg and of course Tallinn, Estonia. Their 10-person team has two Ph.D.s, six Master’s degrees, and 13 spoken languages among them.



Please enter your comment!
Please enter your name here