The coronavirus has sparked a revolution in education, pushing schools and institutions online and driving new demand for e-learning apps.
One among them is ELSA, an artificial intelligence (AI)-powered language platform designed to help non-native English learners improve their speech and pronunciation via short, app-based lessons.
Under the pandemic, the Google-backed company — which uses machine learning to train spoken English specifically — has hit around 11 million users and tapped new markets as international lockdowns have prompted a new need for tech-based learning solutions.
But when Vietnamese entrepreneur Vu Van founded the company in 2015, it was out of a whole different necessity.
A billion-person problem
Van still remembers struggling to find her voice.
Having relocated to the U.S. from her native Vietnam some years before, first for study then for work, she frequently found herself lacking the confidence to speak out, despite being fluent in English.
It was a problem shared by her non-native peers. Concerns over mispronunciation held them back in her Stanford MBA class and, later, management consulting work, often leading them to be overlooked or, worse still, said Van, mistrusted.
To speak confidently and fluently so that other people can understand you, that can be fixed.
Vu Van, CO-FOUNDER AND CEO, ELSA
And if it was an issue for them, it was an issue for many others, too. Of the approximately 1.5 billion English speakers globally, the World Economic Forum estimates over 1 billion are non-native or learning English as a second language.
So Van decided to do something about it, dreaming up a tech-enabled tool that could accurately detect users’ broken English and provide easy-to-follow solutions at a fraction of the cost of a tutor.
“To get a perfect American accent or British accent, that’s very hard. But to speak confidently and fluently so that other people can understand you, that can be fixed. And if there’s a lot of benefit in doing so, then why not?” Van Vu said.
English language app ELSA provides users with feedback on they perfect their speaking and pronunciation.
Finding a co-founder
Still, with no AI or machine learning experience of her own, Van knew she had her work cut out making her vision a reality.
Having quit her consulting job, she spent the next six months searching for a technical co-founder, speaking to “basically every AI voice recognition expert in the Bay area” to gauge their interest and gain their insights.
“My approach was very simple: Every day I just need to talk to five people. I don’t care who they are as long as I can get connections and then those five people will introduce me to another five people,” she said.
Van’s search eventually led her to Germany, then-host to the world’s largest voice recognition technology conference, after a technical professor advised her “if you don’t find anybody there, then you might as well shut the company down.”
A meeting of minds
There, among a gathering of 3,000 eIt was a process that would require total honesty, with “all the toughest conversations being had early on,” such as agreeing on salaries and equity splits, acknowledged Van, who had collated a checklist of questions to ask with the help of her fellow founder friends.
“We said if we don’t kill each other by the end of the three months then I think we might be okay,” she recalled.
Going viral
But the high stakes approach paid off. With Anguera in place as co-founder and chief technology officer, the pair immediately set to work building a prototype; inputting data from non-native English speakers and benchmarking it against standard American English.
For Van, that meant hitting the ground in her native Vietnam to help train the AI against a broad set of non-native English speakers, from bus drivers to boardroom executives.
A meeting of minds
There, among a gathering of 3,000 experts, Van met Xavier Anguera, a top scientist who, as she put it, “had been in research for way too long and was itching for that impact.”
Within weeks, he’d agreed to join her, temporarily leaving his family in Portugal and relocating to Van’s “tiny” San Francisco apartment to stress test the partnership and build out their idea.
It was a process that would require total honesty, with “all the toughest conversations being had early on,” such as agreeing on salaries and equity splits, acknowledged Van, who had collated a checklist of questions to ask with the help of her fellow founder friends.
“We said if we don’t kill each other by the end of the three months then I think we might be okay,” she recalled.
Going viral
But the high stakes approach paid off. With Anguera in place as co-founder and chief technology officer, the pair immediately set to work building a prototype; inputting data from non-native English speakers and benchmarking it against standard American English.
For Van, that meant hitting the ground in her native Vietnam to help train the AI against a broad set of non-native English speakers, from bus drivers to boardroom executives.
However, the true turning point came a few months later, when ELSA won South by Southwest’s 2016 start-up competition, causing the app to go viral, amassing 30,000 users within 24 hours, and granting the team access to user data from across the world.
“The goal at the beginning was collecting data, so the faster we can get there the faster we can train our AI,” said Van.
Winning Google’s backing
With an international dataset in place to educate the technology on a range of non-native English accents, from India to Spain, the wheels were set in motion.
Gradient Ventures, Google’s AI fund, leads $7M investment in English learning app Elsa
Shortly afterward, having relied on their own savings for around six months, Van and Anguera secured an initial seed investment to grow the business. By early 2018, with a growing team and several million users across 100 countries, ELSA secured $3.2 million in funding, including from Southeast Asia-focused venture capital fund Monk’s Hill Ventures.
We were very inspired by Vu and Xavier’s conviction in solving a real problem.
Peng T. Ong, CO-FOUNDER AND MANAGING PARTNER, MONK’S HILL VENTURES
“ELSA was one of our first investments in Vietnam where we were very inspired by Vu and Xavier’s conviction in solving a real problem for over 1.5 billion English learners,” Monk’s Hill Ventures’ co-founder and managing partner Peng T. Ong told CNBC Make It via email.
That vote of confidence was bolstered in 2019, when backing from Google’s AI-focused Gradient Ventures took total funding raised to more than $12 million and granted ELSA access to Google’s team of technical staff to help build out its backend infrastructure.
Surging under the pandemic
The boost came just months before the coronavirus pandemic overturned education and supercharged the growth of online tools.
ELSA — which operates a freemium model that gives users full access to over 1,000 courses for around $3-$6 per month, depending on their package — has since seen user numbers surge “three-to-four times” on a monthly basis, according to Van.
That growth is not only from ELSA’s typical users, but also from schools and businesses adapting to new ways of teaching. The company has now partnered with dozens of schools and enterprises across Vietnam and India, as well as Brazil and Ukraine, as it expands into the business-to-business market (B2B).
“Covid really opened up a segment that is new for us,” said Van. “There’s a paradigm shift among parents that there’s a different way of learning. Instead of always having to send their kids to a language learning center or a school, they can rely on technology. We ride on the benefit of that.”
Building for the future
As the pandemic rolls on, that demand is likely to continue. “In today’s world, fluent English is considered an asset for greater economic opportunity and we expect to see the continuing growth of edutech — partly accelerated by the pandemic — in Southeast Asia with more entrepreneurs bringing education innovation through technology,” said Ong.
Van said that means further fundraising will be “on the horizon” soon, as the company looks to bolster its existing teams in San Francisco, Vietnam, India and Japan, while setting its sights on new markets like Brazil and South Korea.
The new mother also said ELSA is exploring more product innovations, such as continuous monitoring, that would allow the app to give feedback reports based on conversations had throughout the day. Such additions, she noted, will need to closely adhere to data privacy rules.
“2020 has been a crazy year, but I think we have done well and we’re excited for what’s in store for 2021,” she said.
By Karen Gilchrist