Sameer Shariff - Breaking Language Barriers - [Founder’s Field Guide, EP. 46 ]
What this covers
Sameer Shariff, co-founder and CEO of Cambly, tells the company's story in conversation with the host, starting from his time at Google. Cambly began when its founders, who had learned languages in classrooms, noticed that immersion with native speakers taught faster and was more rewarding. They asked why that couldn't be delivered from home. The episode follows from there: the learning case for one-on-one tutoring, with a reference to Benjamin Bloom's two-sigma finding, and the economic case that English gates jobs, migration, and most of the internet.
The middle of the conversation is mostly marketplace mechanics. Shariff explains why an on-demand service needed paid tutors sitting idle at first, so it would feel alive. Matching on occupation, personality, and timing is high-cardinality, so it needs deep liquidity. He also covers the early pricing, where a pay-per-minute model failed because it made learners weigh every minute. It gave way to a use-it-or-lose-it daily subscription. Quality control and moderation also get a segment.
The final stretch covers the company's international growth and where it goes next. Each market opens with one local country manager, beginning with Korea, in a scrappy launch, and the app, support, and marketing are localized. Shariff also describes how hard fundraising was, since investors didn't see the problem daily, and argues for staying lean and raising when it is easy. He closes on an 11-star vision of personalized tutor matches and curriculum, plus his view of hiring people better than oneself and giving them autonomy.
Shariff argues that English learning is a large, overlooked market because English speakers cannot see the access gap, and that a one-on-one, on-demand, two-sided marketplace with subscription pricing solves the missing 'practice speaking with a real person' step that classrooms and apps never provided.
- Most of the roughly 6 billion non-English speakers are invisible to English speakers, so the problem goes unseen
- Subscription 'use it or lose it' pricing builds the daily regular practice that drives fluency
- Hiring local country managers and localizing every touchpoint lets a single product feel native in each market
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
One-on-one conversation with someone more knowledgeable learns fastest because it skips content the learner already knows and stays at the edge of what they do not yet know.
“We don't waste any time talking about stuff I already know. You're always at the fringe, at the edge of your knowledge”
Leaders should hire people better than themselves at the task and give them full ownership and autonomy, because micromanaging executors consumes the leader's cognitive load and frustrates both sides.
“What you don't want is to hire people where you're still overseeing all the work and still giving all of the direction on what to do and then they're executing, because that takes a ton of cognitive load for you.”
Immersing oneself in an environment where only the target language is spoken is far more effective for learning a language than classroom study, because it forces the learner to come out of their shell and practice speaking, and is more rewarding.
“what I found is that immersing myself in the environment was a far more effective way to learn the language”
Learning English unlocks economic opportunity such as jobs, promotions, and the ability to move somewhere with a better life, and it also gates access to the internet, since the vast majority of online information is in English.
“the vast majority of the internet is in English. If you don't know English, all of a sudden, the world of information you can play in is much smaller.”
Bloom's 2 Sigma study found one-on-one tutoring produces outcomes about two standard deviations better than classroom instruction, and Cambly is attempting to scale that model, which was previously academic.
“it's a study that was done decades ago that basically compared one-on-one tutoring to learning in a classroom environment”
The company's first hire was a Korea country manager, because a founder could spend ten years learning Korean language and culture and still not match someone who grew up there, so filling skill gaps with local hires was the better choice.
“This is a case where I could spend the next 10 years trying to learn the Korean language and learn about Korean culture and still not be anywhere close to someone who grew up there.”
To make a 24/7 on-demand service feel alive, the founders paid tutors out of pocket to sit idle and wait for calls, because a marketplace needs supply available before demand arrives.
“we basically effectively had to just pay people to sit around and wait on the off chance that someone called in and wanted to talk to a native English speaker”
The problem of not speaking English is enormous, with about 6 billion of 7.5 billion people lacking English and 1.5 billion actively trying to learn it, yet English speakers overlook it because their social networks are also English-speaking.
“there's about 7.5 billion people in the world right now, six billion of them don't speak English. In the broadest sense, that is the market. And by the way, on those six billion people, 1.5 billion people are actively trying to learn English right now.”
Students pay a monthly subscription with a fixed daily minute allowance, use-it-or-lose-it, which encourages regular 15-minute daily practice that is more effective than binge sessions, and the design teaches usage habits.
“Say you pay 100 bucks a month and you get 15 minutes a day daily. And it's use it or lose it”
Cambly launches each new market with a single country manager, often a local resident, who runs a scrappy go-to-market playbook and then hires a team, which is more effective than trying to learn a culture from outside.
“I could spend the next 10 years trying to learn the Korean language and learn about Korean culture and still not be anywhere close to someone who grew up there.”
Building a lean team that scaled the business with little capital created a strong trajectory, and the painful fundraising struggle ultimately put the company on a better path.
“lack of capital, it can often be a strength. We've built a really lean team that's scaled the big business and we've done it with like very little capital.”
Good matching requires enough scale and liquidity within each vertical because the number of relevant dimensions, such as occupation and personality, is high-cardinality, so a single good match requires a large pool.
“you actually need enough scale in order to deliver on really good matches and you need enough liquidity, not overall in your community, but within each vertical.”
Software and classroom tools teach fundamentals but never reach fluency, because learners study for years without ever practicing speaking with a native speaker, and the workbook is easier than arranging that conversation.
“it's much easier to go pick up a workbook or a book and study from that than it is to snap your fingers and have an English speaker in front of you.”
A pay-per-minute model failed because it did not teach students to practice regularly, and per-minute cost made learners evaluate each minute, which discouraged the usage the product needs.
“it makes you evaluate whether every minute is worth it.”
Cambly originated when co-founders who had learned languages in classrooms realized that immersion with native speakers, not classroom study, produced faster and more rewarding language gains, and then asked why that experience could not be delivered from home.
“what I found is that immersing myself in the environment was a far more effective way to learn the language.”
To match well, the platform must collect structured data on students and tutors across dimensions such as personality, interests, and occupation, plus timing, because matching two people is complex.
“what do you need to do to that is effectively collect structured data on both sides of the marketplace so then you can match in some of those dimensions”
Cambly localizes the app, support, and marketing, so customers in Saudi Arabia and Korea think it is a local company, because every touchpoint is culturally aware rather than just translated.
“if you go to Saudi Arabia and you ask our customers about Cambly, they think we're a Saudi Arabian company.”
Paying a stranger for conversation was a strange new idea that Cambly had to normalize, as Airbnb normalized sleeping in strangers' homes, and winning people over takes time.
“the idea that people would pay to just talk to a friendly native English speaker, that was a novel idea, that was a category that we created”
Fundraising goes well when the business needs less capital and has traction, so the advice is to stay lean, grow the business, and raise when it is easy rather than when it is hard.
“the less you need the money, the easier it'll be to raise. And so I'm always a fan of being lean, focusing a lot on just growing the business and keeping your head down and raising when it's easy not when it's hard.”
New markets are launched by hiring a single general manager with a scrappy, analytical, entrepreneurial profile who owns the full local experience and builds a brand from scratch, which is similar to launching a new company but with the product and tutor supply already in place.
“We've learned that you want someone who's like scrappy entrepreneurial, highly analytical.”
Matching a tutor to a student is harder than matching on attributes alone because a student may want a session now, so the platform must have the right match available at that moment, which requires far more supply than matching alone would suggest.
“The student is potentially excited to have a conversation right now, there's also a temporal component, can we deliver this perfect match to you?”
The long-term vision is to deliver a personalized match with a tutor who shares interests, personality, style, and occupation, along with personalized curriculum, rather than just an instant native speaker.
“we're not just going to get you a friendly native English speaker, we're going to get you like perfect person for you.”
Cambly struggled to raise money because investors did not see the problem every day and did not understand it well, so the company had to hit bigger milestones before later rounds.
“it was not a problem that people saw every day and understood super well”
Most of the internet is in English, so non-English speakers have a much smaller accessible world of information, even when resources like a business podcast seem universally available.
“the vast majority of the internet is in English. If you don't know English, all of a sudden, the world of information you can play in is much smaller.”
The smartphone, with its front-facing camera and fast connectivity, made instant video chat with someone across the world accessible to everyone, which is the enabling technology for the market.
“it was really the advent of the smartphone, this basically connected device with a front-facing camera that's perfect for video chat.”
The first major challenge for a technology company is getting a product that works, which Cambly achieved by launching a rough product early and iterating until it delivered value.
“we put a product out there when it was pretty rough and had to iterate a bunch in order to get something that was actually valuable and worked.”
Cambly followed Airbnb closely because both were international cross-border marketplaces creating new behaviors and facing skepticism, and Paul Graham compared the two companies during Y Combinator.
“I remember Paul Graham making that comparison a few times during our course there.”
Cambly moderates by recording conversations so students can review them and flagged conversations can be investigated, because bad actors are rare but inevitable at scale.
“we record conversations and they can go and watch them and review the discussions they've had. And if a conversation gets flagged”
Cambly is a two-sided marketplace in which learners on one side are matched to native English speakers on the other, creating value on both sides because those two populations are geographically far apart.
“On one side, you've got tutors, on the other side, you've got the students.”
One-on-one connections between American tutors and learners in places like Saudi Arabia short-circuit media-filtered views of other countries, revealing that people are more similar than different.
“If you think about the ways you learn about various parts of the world, it's largely filtered through the media and how the media portrays that. I think what's cool about Cambly is we short circuit that”
As the company scaled, Sameer's role shifted from writing code and doing the work himself to building the right team, which required learning finance, pitching, and fundraising and letting go of tasks.
“My role has gone from doing a lot of the work myself to focusing less on product building and more on company building.”
Sameer's parents immigrated and sacrificed risk-taking to provide opportunity for him, and this provided the safety net that makes a high-risk startup path possible.
“it's obviously a high-risk path, I don't think I wouldn't have necessarily been able to do it without a safety net.”
Cambly collects subscription revenue from students and pays tutors a fixed rate per time spent speaking, so revenue collection and payout are decoupled in time and structure.
“it's a interesting marketplace because it decouples the way we like collect revenue and then pay it out.”
Cambly's success across countries depends on population size, how much a country needs English and how well its people speak it, and purchasing power, because the one-on-one product is not cheap.
“purchasing power is pretty important as well because Cambly is not super cheap.”
Patrick proposes that marketplaces rest on four core functions: setting rules and standards, building core tools and services such as matchmaking, and audience building, which Sameer then uses as the frame for the rest of the discussion.
“One of the models I always try to think about for marketplaces is just these four core functions.”
The best country general managers are scrappy entrepreneurs who are highly analytical, fit the culture, and can grow into managers and leaders, since they will want to build teams once they succeed.
“We've learned that you want someone who's like scrappy entrepreneurial, highly analytical.”
Cambly layered messaging, translation, student profiles, and curriculum such as business English on top of the core video chat to make learning effective beyond a FaceTime-like call.
“The atomic unit or core part of the experience is just this video chat, similar to FaceTime. What we've done over the years is layered on a bunch of other technology on top of that”
Katrina Lake of Stitch Fix is a role model whose experience raising tens of millions before IPO and building a strong early team resonated with Cambly's early challenges.
“Katrina Lake at Stitch Fix. ... she raised something like 50 or $60 million before the IPO”
Learners who need English for professional settings require role-specific language and terminology, so courses must be tailored to occupations rather than general conversation.
“if you're trying to learn English for some professional setting, then zooming in on the specific language and terminology you need to use for that role is super important.”
Language learners want a regular person to talk with, not a formal lesson, so the product was designed like striking up conversations while traveling rather than like a class.
“what we wanted wasn't necessarily a formal class or a formal lesson, what we wanted a regular person to talk to”
In the earliest days the founders were the only tutors, so they spent their days building product while answering calls, which was a symbiotic way to learn customer needs.
“my co-founder and I were the only two tutors on the platform, and so we would spend our days basically writing code and building the product”
Like Uber or Lyft, a simple button hides substantial backend work, including predicting demand and ensuring enough tutors exist at any hour to fulfill it.
“We had to build a lot of technology to, for example, predict demand, like make sure you always have enough supply to fulfill whatever demand you have any hour of any day.”
The one-on-one expert-connection model, as seen in profitable expert network businesses, can be powerful in many contexts beyond language learning, such as on-demand doctor consultations over video.
“there's other companies out there that have zoomed in certain categories like getting a doctor on demand, for example, over video chat.”
The underlying technology of on-demand one-on-one video matching could extend to other verticals such as on-demand doctor consultations, though Cambly is focused on English education because the English market is large enough.
“I think the technology and the product absolutely extends to other arenas. And there's other companies out there that have zoomed in certain categories like getting a doctor on demand”
Early customers came through App Store discovery and word of mouth, driven by a strong first-session experience in which a user is connected to a native speaker within seconds.
“A lot of our students from the early days came through App Store discovery and word of mouth.”
Cambly is deliberately staying in English education because the market, six billion non-speakers and 1.5 billion learners, is large enough that the company is far from exhausting it.
“when you think about the size of the market... we are stuck in that market.”
Cambly chose to focus on which learner segment to serve first based on where the experience of live conversation creates the most value, rather than on a fixed level-based curriculum, though the exact segment choice is treated as a strategic decision.
“what we wanted wasn't necessarily a formal class or a formal lesson, what we wanted a regular person to talk to”
A Brazilian learner who had never met a native English speaker went from zero exposure to having an American available in his pocket, and this dramatic before-and-after contrast drove sharing.
“he's gone from never meeting a native English speaker in his life to quite literally having an American in his pocket anytime he wants.”