Praktika

Praktika picked Brazil from English proficiency data, not founder intuition. The English First report on English penetration showed Brazil as one of the least penetrated markets, with a large income gain from speaking English and a population in the hundreds of millions.

The team worried about weak purchasing power but judged the fundamentals strong enough. The app launched worldwide but marketed only to Brazil and taught only English, and Brazil became the source of the go-to-market claims later used globally.

Sub Club▶ 10:03

Sell the career outcome, not the lessons, so the price feels justified. In a market with low purchasing power where people rarely pay for consumer subscriptions, Praktika sold a route to a better job and got people paying about $100 a year. Claims were framed against the cost of a tutor, which produced the "50X cheaper than a tutor" line.

The product itself teaches language, not job search, so the job promise lives in the marketing. "We didn't sell English to them. We sold them an opportunity to get a better job." Host Jacob Eiting adds that products that help people get richer are an easier sell than hobbies.

Sub Club▶ 12:45

Recruit creators with high reach and small followings, then leave them alone. Adam Turaev filtered for creators with under 5,000 followers but tens of millions of views, reasoning that the algorithm had already picked them.

His rule: "Do not tell influencers how to do their job."

Sub Club

One correction video Adam disliked took ARR from $1M to about $12M. Creator Gustavo's video of a coach correcting a learner's mistakes got 10M views each on Instagram, TikTok and YouTube, and ARR went from $1M to $10-12M in about a month.

Praktika immediately copied the format with other creators and countries, varying the angles, on a tiny ad budget, and it became the standard for language app videos. No breakdown of paid spend, other creators' results or retention is given, so the single-video cause is unproven.

Sub Club▶ 40:19

Staff the creator program with former creators, not managers. The influencer lead was a creator in Poland making Praktika videos before being hired, and country coaches are also ex-creators, each rotating 20 to 30 creators across a program of over 100.

Leads and coaches are in-house while creators are contractors. "They know what good content is, and that's the key skill you need to have."

Sub Club▶ 43:58

Keep organic creators and paid ads as separate streams. Creators invent their own angles, and the paid team only reproduces videos that go viral, using its own internal influencers or contractors.

This keeps fresh ideas coming when the in-house team runs out of angles. "Those two are separate streams. We don't mix them up."

Sub Club▶ 48:02

Pay creators a fixed fee plus a views bonus, flexing for top talent. The standard deal is a fixed part plus a bonus for views, and coaches can change the structure to land a strong creator.

The program runs on SideShift with Praktika's own add-ons for analytics and reporting. Cost per creator, payout sizes and how many creators produce winners are not disclosed.

Sub Club▶ 46:41

Let growth content be unhinged, but keep product rules on one page. Cursing avatars went viral in upper-funnel ads, but users inside the app disliked them and did not understand why their coach was swearing; Adam calls it the biggest failure of the year.

It clashed with Praktika's positioning as an AI coach, not a companion. Product boundaries now fit on a one-page sheet, while creators keep almost unlimited freedom.

Sub Club▶ 50:36

Years of enterprise chatbots taught the team conversation design. Three school friends founded Cleverbots in 2017, inspired by Telegram's chatbot API, automating up to 60% of support and HR tickets for large international companies.

Building hundreds of chatbots showed them what users expect from a reply and where they churn. The pivot followed a pain point the founders felt as non-native English speakers, not the biggest market.

Sub Club

Ship scripted versions early and learn users while the models catch up. Prototyping began in 2021 on the earliest GPT API, a 2022 web version used a Synthesia avatar and hundreds of copywriter-built conversation trees, and a rough app launched in mid-2023.

Hundreds of 2022 customer interviews say what users say today, so go-to-market and user knowledge were built before the models were ready. Praktika reached $50M ARR with almost half a million paying subscribers. "Build on the edge of technology."

Sub Club▶ 14:07

Prioritize quality over inference cost because model prices keep falling. Adam saw a new model launch 50% cheaper, then another 50% cheaper two or three weeks later, so he optimizes for quality and lets costs fall.

Praktika is profitable with about 40 people and more than $1M revenue per employee. "You can focus on quality and be efficient."

Sub Club▶ 55:13

Animated avatars exist to lower the fear of speaking, not for marketing. Fear was one of the most clicked options in onboarding, so avatars simulate being watched by someone; even the 2022 version had an avatar reacting to users.

A Pixar-style look avoids the uncanny valley, since avatars need not be perfectly human. "It was never with marketing or distribution in mind."

Sub Club▶ 30:51

A structured coach beats a general voice model's sparring partner. Praktika scores learners on the CEFR scale (A1, B1) in onboarding and guides them toward a goal in a judgment-free space, which Adam says separates it from ChatGPT voice.

The job to be done, a better job through English, drives product choices, so context and taste matter more than the model. "We're not offering just an AI companion, it's an AI coach, AI tutor."

Sub Club▶ 23:09

Let the AI coach set things like reminders through conversation. Instead of a settings screen, the tutor asks whether the learner wants a daily reminder and sets it for them.

Adam sees agents handling more of the experience this way in future.

Sub Club▶ 54:50

Real-time translation worries investors, not learners. Adam hears the concern in investor talks but never in customer interviews; an American in Italy used his Praktika coach to talk to waiters instead of Google Translate.

Translation covers transactions, but belonging in a new culture requires speaking the language, so learners still want to speak themselves.

Sub Club

Moving from Flutter to native iOS lifted conversion and engagement. Praktika expected only to avoid losing metrics, but conversion and engagement improved across the board, and Adam calls it the biggest win of the past year.

The app became more stable and better suited to the device. He would go iOS native first if starting again, then build Android.

Sub Club▶ 58:04