There's something almost cheeky about teaching English through WhatsApp. The platform that most of us use for family photos and work emergencies has become the classroom for BeConfident, a São Paulo startup that just landed R$85 million—about $17 million—in a Series A round led by Prosus Ventures.
Founded barely two years ago, BeConfident has managed what many edtech ventures spend years pursuing: actual revenue at scale. The numbers tell a story. Three million users. Some 160,000 paying students scattered across more than 62 countries. Revenue hit R$60 million in 2025, a figure that would make most early-stage startups envious.
But here's where things get interesting, or perhaps reckless, depending on your appetite for risk. The company's founders—CEO Robson Amorim, CTO Felipe Tiozo, CPO Felipe Silva, and CMO Luan Cavallaro—are projecting R$300 million in revenue for 2026. That's five times their current run rate. In twelve months.
Even by the hyperbolic standards of venture-backed growth targets, it's aggressive.
The Money Trail
Prosus Ventures didn't write that check alone. Rethink, Alexia Ventures, and Endeavor joined the round, collectively valuing BeConfident at roughly R$530 million, or around $100 million. That's a meaningful leap from the company's genesis: a modest $500,000 pre-seed from Latitud and a constellation of Brazilian tech's elite angels—André Penha of QuintoAndar, Manoel Lemos of Redpoint, Doug Scherrer from Nubank.
Early 2025 brought another inflection point. Matthew Brezina's Ford Street Ventures led a strategic round that pulled in Grão, Entrypoint, and Sun Moritz Family Office, nudging the valuation to R$100 million. Now, just months later, that figure has quintupled.
The trajectory suggests either profound investor conviction or the kind of momentum that can't be ignored in a frothy market—possibly both.
Conversation Over Gamification
BeConfident's product thesis diverges from what has become the dominant paradigm in language learning apps. While Duolingo turned language acquisition into a game—complete with streaks, cartoon mascots, and dopamine-triggering rewards—BeConfident opted for something less polished but potentially more practical: conversation.
Users interact with AI tutors via messaging, voice, and video. Not bite-sized lessons. Not fill-in-the-blank exercises. Just continuous practice speaking and writing, the kind of immersion that traditional classroom instruction promises but rarely delivers at scale.
Whether this approach resonates long-term remains an open question. Gamification works because it hijacks our psychology; conversation requires genuine effort. But for students serious about fluency rather than maintaining a streak, the trade-off might be worthwhile.
International Ambitions (and Dependencies)

The Series A capital will bankroll an expansion that, if successful, would fundamentally reshape BeConfident's user base. The company is targeting one million new paying customers over the next 24 months, with a striking stipulation: 90% must come from markets outside Brazil—specifically the United States, Europe, and Asia.
That's not wishful thinking entirely disconnected from reality. By mid-2025, the U.S. already represented roughly 45% of BeConfident's international revenue. Early traction, yes, but translating that into sustained growth in markets already saturated with English learning options will require more than a clever WhatsApp integration.
Prosus, for its part, believes the addressable market justifies the bet. The venture arm estimates the global language learning industry at $65 billion, with concentration in India, Latin America, and parts of Europe—regions where English proficiency correlates directly with economic mobility.
Building the AI Layer
BeConfident isn't standing still on product. The company launched BeConfident Labs, a research unit exploring personalized learning models with partnerships that include Stanford-affiliated researchers. The ambition extends beyond generic AI tutors into something more specialized: a marketplace for expert and creator-specific AI avatars.
Imagine learning business English from an AI modeled after a Wall Street analyst, or practicing conversational Portuguese with a virtual Rio tour guide. It's the kind of vertical specialization that could differentiate BeConfident if executed well—or fragment focus if spread too thin.
The Math Problem

Here's what keeps me up at night about BeConfident's trajectory, or would if I had money in the deal: the math.
Hitting R$300 million in revenue next year means adding roughly R$240 million—about $48 million—in net new annual recurring revenue. With 160,000 paying students currently generating R$60 million, the average revenue per user sits around R$375 annually, or about $75.
To reach R$300 million without dramatically increasing prices (difficult in price-sensitive markets like Brazil and India), BeConfident would need somewhere in the neighborhood of 800,000 paying users. That's quintupling the paying customer base while simultaneously expanding into new geographies where customer acquisition costs tend to spike.
Not impossible. Duolingo scaled aggressively. Preply found product-market fit. But those companies had years, not months.
Perhaps the founders see something the rest of us don't—a viral coefficient, a distribution channel, or partnerships that haven't been announced. Or perhaps they're banking on the kind of exponential growth that justifies venture capital's existence but rarely materializes quite as projected.
What Actually Matters

Strip away the valuations and revenue targets, and BeConfident represents something worth watching: a genuine attempt to make language learning accessible through infrastructure people already use. WhatsApp has more than 2 billion users globally. Most of them don't open Duolingo.
If BeConfident can meet learners where they already are, using technology sophisticated enough to provide real value but simple enough to disappear into daily habits, the company might just earn its ambitious projections.
If not, well—R$85 million buys quite a bit of runway to figure it out.
