The dawn of turbocharged language learning

Dr. Arnar Jensson, CEO at Cooori

Dr. Arnar Jensson, CEO at Cooori

For decades marketing firms have promised speedy if not effortless ways to learn a second language. None have worked as hoped. However, rapid advances in machine learning algorithms — the same technology behind self-driving cars, robots and speech recognition known as ‘artificial intelligence’ — may well transform language learning. The disruptive technology promises to optimize language learning through personalization.

At the forefront of the AI language learning revolution in Tokyo is Arnar Jensson. Jensson aims to make language learning faster and easier through his company, Cooori. The firm claims some of Japan’s largest companies as clients, including Japan Airlines and The Dai-ichi Life Insurance Company.

Jensson first came to Japan in 2003 to earn his PhD in computer science at the Tokyo Institute of Technology on a full scholarship from the Japanese government. He studied AI speech recognition under a leading professor and in a world-class laboratory. On the university’s supercomputer, he and other PhD students ran programs in a day that would have taken a year on a personal computer. Arguably, they pushed the boundaries of speech recognition when AI seemed to others like science fiction.

He also studied Japanese for one or two hours each day on the side, taking group classes, private lessons, computer courses (like Rosetta Stone) and so forth. Whatever methods he tried, he could not learn the language. When taking group classes, for example, other students would brush through the course material when he fell behind. Three weeks into the lessons, he had to restart from the beginning.

After graduating with his PhD, Jensson interviewed at a few famous Japanese firms. But what he really wanted to do was to help millions of ordinary people learn languages faster and more easily. Without money or business experience he started Cooori in April 2009. “I knew zero,” he says.

To save on costs, he made Yoyogi Park his office. Every day Jensson sat in the park for 4 to 6 hours with his laptop planning the business. In the evenings when his friends went out drinking, he would arrive late at the bar to order water. For nourishment, he survived on rice and potatoes. “It was tough, but necessary,” he says, adding “Otherwise, I could not have launched the business.”

Jensson began looking for finance just after the Lehman crisis when there was no venture activity in Japan. So he flew to Reykjavik to present before VCs in his home country at a pitching contest. His was selected as one of the top 10 startup ideas out of 200 presenting. Even so, Icelandic VCs were hardly rushing to invest in the aftermath of their country’s bankruptcy.

It wasn’t until the end of 2010 that Jensson started getting strong interest. Still investors asked, “Why are you talking to VCs in Iceland when you live in Japan?” Jensson replied, “I am an Icelander. We are going to teach many people foreign languages. That will be good for the world and for your karma.” Unconvinced, they kept asking for more information. He provided it, long after he felt there was a genuine need.

Jensson imagines that investors gave him $1 million they did in February 2011 only after saying to themselves, “We tried to get rid of him, but he’s still around. This guy has persistence.” For two years up until then Cooori had been self-financed. Now Jensson and his development team could take salaries.

The firm went ‘global’ right from the start, targeting the $56.3 billion worldwide language market. They supported multiple learning language pairs, servicing the B2C, B2B, even B2A (‘A’ for ‘Anybody’) markets. After acquiring hundreds of thousands of users of all kinds — each expecting and demanding something slightly different — Cooori was flooded by all kinds of requests they couldn’t properly manage. “It was awful,” reports Jensson.

A cash flow crisis in late spring 2015 sparked a strategic rethink. Instead of going after the entire market, they thought about better servicing a narrower, profitable niche. Two years earlier Cooori had nabbed Japan Airlines as a client, after winning a startup competition sponsored by them. Unlike B2C users who were only willing to pay a couple dollars for a language learning app, B2B client JAL more highly valued Cooori’s ongoing service.

Generally speaking, HR departments liked Cooori because it is more cost-effective than sending employees to English conversation schools. They can also better monitor their educational return on investment. Jensson reasoned that if Cooori focused on servicing the Japan B2B enterprise market, it might just get them out of a jam.

Time was running out. “With only a couple weeks of cash flow left, we were worrying how we could get out of this mess,” Jensson recounts. He and his staff didn’t know how to pitch current investors. “So your previous strategy wasn’t good enough?” they asked during a make or break meeting. “No, not good Eddy,” Jensson replied adding, “Work with me on this one.” Thankfully, VCs topped the firm up with $4 million invested so far.

Cooori had to apologize to most customers, advising their service would end after fulfilling current contracts. The firm then focused on better servicing only large Japanese companies. “Had we not pivoted, I’m sure Cooori would have gone bankrupt,” he admits.

Jensson compares Cooori’s enterprise product to an 800 hp rally car with a seat belt, but no windshield, air-conditioning or leather seats. He claims language students learn “ten times faster” with Cooori than by any other method. “Like a rally car, you press the pedal and you feel the g-force. Once you’ve tried it, you’ll never go back to a Volvo,” he says.

Now click here to read our next article, ‘Adaptive language learning at the starting gate‘.

Beacon Reports reveals Japan through the lens of thought leaders. Subscribe free!

Share articleShare on LinkedInShare on FacebookTweet about this on TwitterEmail this to someone
,

Leave a Reply

Your email address will not be published. Required fields are marked *