Daybreak - Can Duolingo keep India speaking when AI can translate everything?

Episode Date: December 25, 2025

AI is changing how people learn languages and India is where the shift is showing up first. Duolingo has scale here but very little conversion. At the same time AI tools now offer practice, feedback, and even conversation for free, while Indian platforms focus on jobs, exams, and real outcomes. In this episode, we look at how language learning is being reshaped in India, why translation is no longer the whole story, and what Duolingo is really defending. Tune in.

Hi, this is Rohan Dharma Kumar. If you've heard any of the Ken's podcasts, you've probably heard me, my interruptions, my analogies, and my contrarian takes on most topics. And you might rightly be wondering why am I interrupting this episode too? It's for a special announcement. For the last few months, I and Sita Raman Ganeshan, my colleague and the Ken's deputy editor, have been working on an ambitious new podcast. It's called Intermission.

We want to tell the secret sauce stories of India's greatest companies. Stories of how they were born, how they fought to survive, how they build their organizations and culture, how they manage to innovate and thrive over decades, and most importantly, how they're poised today. To do that, Sita and I have been reading books, poring over reports, going through financial statements, digging up archives, and talking to dozens of people. And if that wasn't enough, we also decided to throw in video into the mix. Yes, you heard that right. Intermission has also had to find its footing in the world of multi-camera shoots in professional studios, laborious editing, and extensive post-production.

Sita and I are still reeling from the intensity of our first studio recording. Intermission launches on March 23rd. To get alert, as soon as we release our first studio recording, episode, please follow intermission on Spotify and Apple Podcast or subscribe to the Ken's YouTube channel. You can find all of the links at the ken.com slash I am. With that, back to your episode. If you've ever used Duolingo, you know the owl. It's cheerful, but also slightly judgmental. And if you miss a few days, the cartoon bird has a way of making you feel personally responsible for disappointing it. For millions of people,

that owl is the face of language learning. It is a comfortable routine, just a daily tap before bed. And by the numbers, Duolingo has every reason to feel confident right now. In 2025, the company expects to cross a billion dollars in revenue. More than 50 million people use the app every single day. And it is one of the most successful education products ever built. And it wrote the smartphone revolution almost perfectly. But, Here is the part that does not show up in the app store screenshots. One of Duolingo's biggest markets, arguably also its most important one, is also its most challenging one.

India. Millions of Indians use Duolingo every day. It is the company's fifth largest user base and the scale is just enormous. But the number of people who actually pay is quite tiny. So you have this strange situation. Duolingo is everywhere in India, but it is not really winning here. Not yet. And at the same time, the ground beneath language learning itself is starting to shift. AI tools that were never designed to teach languages are suddenly doing it very well.

Local Indian startups are quietly building alternatives that care less about learning streaks and more about outcomes, jobs, exams, immigration and overall confidence. And translation itself is becoming instant and built into the devices that we carry, which raises an interesting question. If machines can translate for us and now even teach us, what exactly is a language learning app selling anymore? And in a market like India, where language is tied to opportunity, not just travel, can Duolingo adapt fast enough to stay relevant. Welcome to Debray, a business podcast from the Ken. I'm your host, Nick Da Sharma, and I don't chase the news cycle. Instead, every day of the week, my colleague Rachel Vargis and I will come to you with one business story that is worth understanding and worth your time.

Today is Friday, the 26th of December. There is a phrase in French that fits this moment quite well. And it goes, a mo more to recule. And it means taking a step back or pausing. You know, looking at the whole picture instead of just reacting to the immediate noise. Duolingo likes to remind you. people that it has navigated a platform shift like this before. When mobile phones overtook desktops, Duolingo bet early and hard on apps.

While older Web First education platform struggled, Duolingo became the most downloaded education app in the world. And that story is true. But Mobile also exposed Duolingo's longest running criticism, that the app was excellent at building habits, but not so good at building actual confidence. People did show up every day and they protected their streaks, but many still hesitated when it came to actually speaking. For example, Somaia's experience captures this perfectly. She's a lawyer in Delhi and Malayalam is her mother tongue. Hindi came later and during COVID she decided to turn to doolingo to polish her Hindi. The app did help. Her vocabulary improved and she enjoyed the structure, the rewards and the feeling of progress.

But pronunciation lagged behind and real-world confidence didn't quite follow. After 250 consecutive days, she decided to stop using the app. She still gets a small thrill when she understands Hindi dialogue in a Bollywood film, but speaking fluently without hesitation still feels like unfinished work for her. Duolingo knows that this gap exists and it's. is now trying to close it. New subscription tiers, a Duolingo score meant to measure more than just streaks, and AI-powered features designed to make practice feel closer to real conversation. All of this is happening while the broader logic of language learning is being questioned.

In the US, the number of people learning languages has dropped sharply over the past decade. In Australia, fewer than one in ten students studies a foreign language. University departments are shutting down in places like South Korea and New Zealand. There are many reasons for this, but AI-powered translation sits uncomfortably close to the top of the list. If translation is instant, accurate and everywhere, why even put in the effort? Now, India has not fully felt this effect yet, and that is why it matters so much. In India, English is not just a hobby. We know this. It's not just a holiday. skill. It functions more like a basic credential, which is like a gateway to education, jobs, immigration,

and upward mobility. So that difference is crucial. And it is why Indian founders saw an opening. Multibhashi was built around real-world outcomes from day one, communication, certification, workplace readiness, less focus on gamified dopamine and more focus on usable skills. Language tutors too are seeing the same thing play out. So as a result, India's language learning market has fractured. You have gamified self-learning apps like Duolingo and AERLARN, tutor-led hybrid platforms like PREPLE and Multibhashi, and now AI-first chatbots that sit somewhere in between.

Cheap, instant, but deeply unsocial. Each model solves a different problem. None of them actually have the complete answer. And sitting beneath all of this is the question that nobody can avoid forever. And it is, if translation becomes truly effortless and instant, does learning even matter? For more on this, stay tuned for the next segment. You see, for years, AI's promise was quite simple when it came to languages. You don't need to learn them anymore because you can just translate everything.

But this year, that promise quietly changed. Every major AI platform added language learning features, guided lessons, vocabulary drills, pronunciation feedback and conversation prompts. These were tools that look, feel and function a lot like paid language apps. For these companies, this expansion is almost frictionless. Their models already know the languages and teaching them is just another interface, which puts pressure on everybody else. Why would anyone pay for a subscription?

Duolingo's leadership is also very aware of this. Their answer is not just better answers because AI already handles that. Their answer to this is structure, habit formation, routine and accountability. The streak, Duolingo thinks, is not just a gimmick. It is a behavioral design. It gets people to do something difficult every day long after motivation fades. And that belief shapes Duolingo's AI strategy as well. The first liver is speed. Before large language models, every course had to be built manually by curriculum experts.

Teaching French to English speakers was a completely different project from teaching French to Hindi speakers. It took Duolingo 12 years to build its first 100 courses. And last year, it built 150. In India, that sort of speed really matters. Duolingo rolled out dozens of new vernacular pathways. which is global languages taught through Hindi, Bengali, Tamil and Telugu. But with speed comes tension. When CEO Louis Vuong-Aarns said Duolingo would gradually stop using contractors for work AI could handle,

the backlash was immediate. Clarifications followed, but the direction was clear. Course creation is now AI first. The second lever is personalisation. Duolingo Max, the company's GPT-powered tier, adds conversational practice through video calls, with in-app characters and guided role play charts. It has helped push revenue per user up globally.

But India is different. Subscription conversation, like I said earlier, is low. Max has not fully been launched here yet, and Duolingo is still experimenting with how to offer AI features at prices that Indian users will actually accept. And the third lever is efficiency. Duolingo runs with fewer than a thousand employees worldwide. is embedded across teams to keep the company lean.

But in India, these advantages blur because premium features don't pay yet, and that opens the door for other models to compete. Which is why Duolingo's real moat in India is not the app at all. It is the Duolingo English test. India is the test's largest market worldwide. The idea is simple and powerful. Teach people English through the app, certify them when it matters for college applications, visas and global opportunities. Learn and get tested in one ecosystem. At the same time, Duolingo is also

hedging its bets. Maths, music, chess, subjects that could matter even if language learning slowly becomes fully commoditized. Competitors are also adapting. Hybrid platforms are leaning into what AI still struggles with, cultural context, tone, confidence and relatability. Tudors talk about moments where a joke or a story or an offhand reference unlocks understanding in a way that no chatbot can manage yet. Everybody is adding AI, but nobody is willing to surrender entirely to it. So if language learning becomes a commodity, India will feel it first. If aspiration continues to outrun translation, India will prove that too. And somewhere in that tension, TuoLingo will find out whether the owl can keep flying in a world

where machines already speak every language. Daybreak is produced from the newsroom of the Ken India's first subscriber-focused business news platform. What you're listening to is just a small sample of a subscriber-only offerings and a full subscription offers daily, long-form

feature stories, newsletters and a whole bunch of premium podcasts. To subscribe, head to the Ken.com and click on the red subscribe button on the top of the website. Today's episode was hosted and produced by my colleague, Snicktha and edited by Rati F.N.