9 Artificial Intelligence Startups in Estonia
At roughly twice the size of New Jersey, Estonia is a small post-Soviet state bordering the Baltic Sea. Its size belies the extent of technological development the country has achieved over the past three decades. In the early 90s, a government of radical reformers overhauled the old macroeconomic policies in favor of new open and liberal ones. A state-driven modernization of infrastructure followed shortly thereafter, beginning with the introduction of a digital landline telephone network, a cutting-edge solution back in the day. In 2007, Estonia became the first country to allow online voting in the general elections, a single milestone in the process of full state digitalization that’s in progress to this day. In 2014, the country introduced an “e-residency” that allows applicants to establish themselves and their business in Estonia (and consequently in the EU) and conduct business from there globally.
Registering a company in Estonia takes a mere five minutes, so it’s no surprise the country ranks third in Europe in “number of startups per capita.” In May 2018, ministers responsible for digital development in the Nordic-Baltic region jointly released a “Declaration on AI in the Nordic-Baltic region” with the goal of “developing and promoting the use of artificial intelligence to serve humans.” The document outlines areas of collaborative development including skills development, data access, development of guidelines and standards, and deregulation. In order to see what’s happening in the world of Estonian AI startups, we took a look through Crunchbase and found nine AI startups with disclosed funding.
|Yanu||Food & Beverage||Tallinn||1.1|
|Hala||Business Process Automation||Harju||0.2|
Founded in 2013, Lingvist has raised $11.2 million to develop a language-learning tool that adapts to students’ level and areas of knowledge, minimizing the time it takes to learn a new language. The startup uses neural networks to assess users’ knowledge and provide them with relevant practice exercises in the form of flashcards and reading, listening, grammar, and speaking challenges. The language app currently supports French, Russian, German, and Spanish courses for beginner and intermediate level learners and aims to bring students “up to speed” in 200 hours. Limited access is free, while a subscription costs $23 per month or $90 for a full year.
Founded in 2015, Veriff has raised $7.8 million to develop an online identity verification tool powered by machine learning. Veriff validates official IDs from more than 190 countries by cross-checking personal details. Innovative ways to accomplish this include verifying selfies against ID photos, or extracting biometric information from video feeds – a feature Veriff claims is one-of-a-kind.
Algorithms also check the “fingerprints” of the login device against multiple parameters including past behavior in order to flag inconsistencies. In case automatic verification fails, the company provides a 24/7 fallback service which can be white-labeled for anyone who wants to integrate advanced authentication methods into their business processes. The solution is priced on a case by case basis and is used by a diverse set of businesses including companies in peer-to-peer lending and scooter sharing.
Founded in 2017, Raison has raised $1.3 million to develop a financial platform incorporating personal finance, payments, and investments into a single app. Raison provides a debit card that supports both normal currencies and cryptocurrencies in parallel, supports both Apple and Samsung pay services, and offers investment opportunities, some of which have been largely the domain of institutional investors. The three investment areas include “blockchain technology,” institutional products like structured notes and Eurobonds, and peer-to-peer lending.
Raison uses AI algorithms in its investment module, employing neural networks to monitor global news outlets, gauge market sentiment, and structure recommended portfolios accordingly. They’ve actually put their entire pitch deck up for anyone to download – something we haven’t seen before – which is equal part alarming and exciting. Their pessimistic forecast calls for 20,000 private persons and 1,000 legal entities on the platform by the end of this year. “Build it and they will come,” comes to mind.
Founded in 2017, MeetFrank has raised $1.1 million to develop a chatbot-based job matching app that employs machine learning. Prospective employees can sign up to the service anonymously and answer a few initial questions to be matched with job opportunities without a CV required. The startup aims to help passive job seekers assess the market without getting involved in an official interview process which would reveal their identities. Frank, the chatbot, manages the conversation in a clear and concise manner while using just enough emojis to make it slightly awkward.
Frank’s user base has grown rapidly following the launch, and it has helped major tech startups like TransferWise and Taxify find talent since then. About half of the available jobs are in IT engineering with the rest made up of supporting roles like sales, customer service, marketing, and project management. The app currently has more than 200,000 users and 2,500 recruiters on its platform. It was co-founded by a man, Kaarel Holm, who was fired by his employer because he was looking for another job. Kind of cheeky if he was doing that on company time, but his employer certainly seems to have lost a talented chap. As Mr. Holm wisely says, “the most talented are not looking for jobs.”
Tallinn startup Yanu is adding bartender to the ever-growing job loss list attributed to AI and robotics. Founded in 2016, they’ve raised $1.1 million to build an AI-powered bar-tending robot that converses with guests and serves 100-150 drinks an hour. The Yanu product is a compact little bar where guests can order using a mobile app or an interactive screen, pay by card or mobile, and receive their drinks from an integrated robotic arm.
The unit can stock up to 2,000 drinks and cocktails in one load, so there’s no need for a refill during the night. It operates quicker and more efficiently than humans, reducing queues at the bar and increasing consumption. The robot is able to freely interact with humans, so it will listen attentively to the great philosophic revelations of drunken customers and maybe throw in an ad or two as well. The startup plans to install the first few robots this year in places where space is at a premium, like airports. Prices for the unit are not public yet, and hopefully, it’s at a price point where the operator won’t need to charge airport prices.
Founded in 2015, Teamscope has raised $920,000 to develop a hiring platform that uses big data analytics and machine learning to understand team requirements and find the best fitting candidates for teams. According to the company, 60% of executives say the biggest challenge in hiring is understanding team culture and 30% of new hires underperform in their first year due to the lack of cultural fit.
Teamscope analyzes existing teams by looking at the strengths, skills, and motivators of members, maps the core values of team culture, and identifies gaps and critical competencies for the team. Candidates are interviewed based on behavioral tests and chosen using algorithms that combine multiple personality assessment frameworks and AI-powered linguistic analytics. The company, present in Tallinn and London, promises to save 12-24 months of salary with each successful new hire that doesn’t need to be replaced in the short run. Sounds like the perfect solution for hiring managers who are incapable of building high-performing teams on their own.
Founded in 2015, Feelingstream has raised $920,000 to develop AI analytics for customer conversations. The startup’s product suite uses Natural Language Processing (NLP) to analyze email, chat, and phone conversations in real-time and come up with conversation sentiment, topics, and categories. The platform offers a number of specific models for automated conversation handling including models detecting sales potential, churn risk, and customer intent, and models developed for specific sectors like banking and telecom.
Feelingstream also allows clients to develop and train their own models using supervised and unsupervised machine learning algorithms. The product suite currently covers English and Nordic languages and is used by public and private institutions in the Nordic region including major banks like Nordea and SEB.
Founded in 2017, Hala has raised $200,000 to develop a platform for business process automation, something that quite a few startups are already doing including unicorns like Automation Anywhere. Dubbed a “digital assistant for business,” Hala automates processes at enterprises running SAP, Oracle, and Microsoft Dynamics, three of the largest enterprise management systems today. The system is equipped with a chatbot that is able to understand and execute commands in natural language, retrieve and visualize data, and modify and configure parts of the enterprise software.
Hala can also analyze existing processes and search for new areas to automate including permission requests, password resets, blocking users, and credit limit checks. According to a company case study, a food industry client using SAP was able to save up to $343,000 a year by automating routine menial tasks which represented 16% of all administrative duties.
Founded in 2015, Examus has raised $100,000 to develop a proctoring solution for online education institutions that monitors the integrity of online exams. Examus’ algorithms use webcams and machine vision to identify students and analyze their behavior during an examination. The software detects violations of exam regulations like when students look away from the screen longer than warranted, change the active window on the screen, leave the desk, or have someone else around assisting them.
Examus assesses baseline behavior of each student, so personal habits and differences in behavior do not influence the outcome. Autonomous monitoring is combined with live supervisors during examinations, who monitor and verify suspected violations and step in as necessary. All exams are recorded and forwarded to universities for later analysis, and the platform can also be integrated with learning management systems to automate post-mortem analytics. Examus is mainly used by Russian academic and corporate institutions. Maybe they ought to just make cheating a criminal offense like the Chinese did.
We originally wanted to title this article “the top 10 AI startups in Estonia,” but that plan hit a brick wall when we only turned up nine Estonian AI startups – total – with disclosed funding in Crunchbase. There are far more than nine AI startups in Estonia, and some like Cora Health reached out to us wanting to be considered for this list. Cora Health uses AI to analyze blood pressure readings, but their policy is not to disclose funding so we couldn’t include them on this list while maintaining our objectivity. That’s fine, as we may look at what they do in a future article. We’re in no position to decide which AI startups are “top,’ and which AI startups are “bottom,” so we usually let VCs do that for us by voting with their wallets. Given Estonia’s track record of solving problems using technology, it’s likely that there are a whole lot more Estonian AI startups hiding out there doing interesting stuff as well.