What follows is not a marketing argument. It is a structural read of two products and the orgs behind them. If you want to know what a company really values, look at what its software optimizes for — not what its homepage claims.
You can learn more about a company’s real priorities by spending an hour with their software than by reading their mission statement.
Conway’s Law: any organization that designs a system will produce a design that mirrors its own communication structure and incentive system. Applied as competitive analysis, it means the product reveals the company.
What follows is not a marketing argument. It is a structural read of two products and the orgs behind them. If you want to know what a company really values, look at what its software optimizes for — not what its homepage claims.
Duolingo’s product is built around streaks, points, leaderboards, and notifications engineered for compulsion. It optimizes relentlessly for Daily Active Users and retention.
The product team reports to people whose success is measured by DAU and retention, not by whether users become conversational.
There is no human in the loop. No teacher. No real conversation partner. No feedback that requires judgment. This signals optimization for scale and margin, not for learning outcomes.
A user who opens the app every day to protect a streak and closes it five minutes later counts the same as one who had a genuine learning breakthrough. The metric cannot tell the difference. If the metric drives decisions, neither can the company.
They are baked into the org’s incentive structure, which means they are permanent as long as Duolingo stays Duolingo.
Deep learning is slow, effortful, and sometimes frustrating. That is the opposite of a streak-protecting dopamine loop. If Duolingo tried to fix this, they would crater their DAU, which would crater their stock price.
They are trapped by their own success.
FluentFox is built around a different belief: the learning outcome is the product, not a proxy metric for it.
Real conversation. Not simulated interaction designed to feel good, but actual back-and-forth that builds the skill.
A student model that evolves. The system knows where you are and what you need next. Lessons are made for you personally, not for the median user.
Feedback that requires judgment. We do not remove the hard parts. The hard parts are the point.
Reading our org backwards: a team that believes fluency is achievable, that AI can accelerate it, and that the measure of success is whether users can actually hold a conversation.
We compete on the thing Duolingo structurally cannot deliver: actual progress toward fluency.
Duolingo is a habit. FluentFox is a result.
The users most ready to hear our message are people who completed courses, maintained streaks for months, and then tried to have a real conversation and discovered they could not.
They already know what gamified learning produces. They have already paid the cost of finding out the hard way.
This is a large and frustrated population. They are looking for something that actually works.
We are that thing.
Does this serve the learning outcome, or does it serve a proxy metric?
Features that create engagement without creating progress are Duolingo features. We do not build those.
Our competitive advantage runs through everything we build. As long as we hold to it, Duolingo cannot touch us.
The moment we drift from it, we become a worse version of them.
Every lesson is made for you, personally.