❓ Problem

Duolingo is not a language company. It is a habit company that happens to sell language.

The proof is in the numbers. It finished 2025 with $1.04 billion in revenue, up 39% year on year, and by mid-2026 was serving 58.7 million daily active users and 12.7 million paying subscribers. None of that was built on a secret method for teaching Spanish. Spanish grammar has been free on the internet for twenty years. It was built on streaks, leaderboards, five-minute sessions and a cartoon owl that makes you feel bad about yourself.

Here is the thing nobody in edtech wants to say out loud. Content was never the bottleneck. Motivation was. Across the entire MOOC era, the completion rate for edX and Coursera courses sat between 5% and 15%, and MIT's analysis of six years of edX data found that just 3.13% of participants finished in 2017-18, with more than half never even starting. Harvard lectures, free, on demand, and almost nobody made it to week three.

Now the twist. Generative AI has taken the cost of building a course to roughly zero, which means the last remaining moat in consumer learning is the loop, not the library. And Duolingo has proved the appetite goes way past languages: it launched Chess in April 2025 and the mode passed one million daily active users within a single quarter. Yet as a public company chasing 100 million DAUs, Duolingo will only ever build the four or five subjects big enough to move the needle. Languages, maths, music, chess. That leaves everything else, which is most of what humans actually want to learn, sitting completely unserved.

Football tactics. Philosophy. Wine. Personal finance. Architecture. Geopolitics. There is no daily five-minute loop for any of it.

✅ Solution

A Duolingo-shaped app where the subject is anything, and the product is the habit.

• You type what you want to understand. "Explain football tactics." "Teach me the basics of philosophy." A short diagnostic conversation works out where you are starting from, how much time you have and why you care, then generates a structured path with levels, units and a visible end point.
• Every day unlocks one five-minute lesson built around doing, not reading. Drag players into position to break a high press. Match the argument to the philosopher. Pick which bottle survives the decade. Interactive challenges, not a chatbot lecture with a quiz stapled on the end.
• The loop does the heavy lifting. Streaks, XP, leagues, hearts, a push notification with a bit of personality. Spaced repetition drags back the concepts you fluffed. Difficulty adapts silently so you are always just uncomfortable enough to keep going.
• Paths are shareable objects. Finish a path and it becomes a link. Creators, coaches, teachers and newsletters can publish their own official course, which turns every enthusiast audience into a distribution channel.

The wedge is deliberately narrow: own one subject where people are already obsessive, nail the daily loop, then let the catalogue widen underneath a habit that already exists.

📊 Key Numbers

Market size

• The global e-learning market runs to roughly $276 billion in 2026 and is forecast to hit $462 billion by 2031. That is the TAM, and it is mostly corporate and academic, so treat it as a ceiling rather than a target.
• The relevant SAM is consumer subscription learning apps. Language apps alone pulled in about $1.11 billion of consumer spend in 2024, with Duolingo taking an estimated 60% of category usage. Non-language self-directed learning has a vastly wider surface area of topics and almost no gamified supply.
• Duolingo's own survey work found that around 80% of its US users were not learning a language at all before they downloaded it. The category does not just capture demand, it manufactures it. Apply that to philosophy, tactics or finance and the addressable pool is far bigger than the number of people currently buying courses.