The four things second-language research says actually work — comprehensible input, explicit grammar, retrieval practice, and real output — each generated and sequenced to your exact level. A weekly article you can actually read, a grammar skill tree built from your level up, an Anki deck and quizzes, and a daily AI tutor that knows what you've been studying.
Stephen Krashen's input hypothesis says you acquire a language by encountering it just slightly above your current level — what he calls i+1. The threshold researchers settle on is roughly 95% known vocabulary: enough familiar words that the unfamiliar ones become legible from context, instead of derailing the sentence.
In practice, almost nothing hits that. LingQ's catalog is unleveled native text — too dense. YouTube channels for learners assume more than you know. Textbooks are dead and generic — the same chapter for every learner since 1985. The result is that "comprehensible input" is usually only partially comprehensible, which means slow gains and a lot of dictionary-tapping.
Frank Lingo generates your weekly article against your actual vocabulary set — the CEFR pool at your level plus the words from your current deck — and measures the unknown-word density before shipping it. If it lands above 5%, it gets rewritten. That's how the input stays comprehensible.
Read the full method →You can't have a conversation in a language you've never read. Reading comes first.The Frank Lingo thesis
A short article in your target language, written from your vocabulary set, with comprehension questions. This is where new words enter in context — sized to your level so the input is actually comprehensible.
Comprehensible input alone isn't enough for grammar — the research is clear on that. So we teach the patterns: a brief explanation, then practice reading and producing each structure, sequenced along a skill tree built from your level up.
A monthly Anki deck of your article's vocabulary with native audio, plus web quizzes. This is what you carry forward — spaced retrieval builds durable memory.
Conversation in your target language, every day, using the same vocabulary and grammar as your reading and lessons — running from day 1, not gated by "readiness."
Plus a monthly progress report with reading comprehension, quiz accuracy, engagement trends, and a recommendation for next month. No streaks, no app, no guilt trips.
30-day free trial — no credit card charged until trial ends. Cancel anytime.
Talking to a bot in a language you don't know is roleplay, not learning. Output practice on top of comprehensible input is fine; output without input isn't. The tutor channel is part of the system, not the system itself — it draws from the same vocabulary set as your weekly article and monthly deck, so what you produce in conversation is what you've actually been studying.
Reading native content is the goal, not the method. Until you have the vocabulary, native text isn't comprehensible — it's just text with a dictionary attached. We generate input sized to your level so reading actually drives acquisition; once you outgrow our level cap, LingQ-style native-content workflows are a great next step.
Every Monday you receive a short article written in your target language, generated against your current vocabulary set (CEFR pool at your level plus your deck words) with comprehension questions. We measure the unknown-word density before it ships and rewrite anything that lands above the 95%-known threshold. Article length scales with your level — short paragraphs at A1, longer pieces at B2 and above.
Yes — Anki is free on all platforms. Each month you'll receive an .apkg file to import. Deck sizes are calibrated to your CEFR level (60 sentences at A1, up to 200 at B2), and each month's import merges into your existing deck — your review history is always preserved. If you don't use Anki, the weekly articles, vocabulary quizzes, and AI tutor still work independently.
The public language picker shows every language where our CEFR vocabulary list is ready. As we band more languages, they appear in the picker without a code deploy. The launch matrix is Mandarin, Swedish, Greek, Spanish, French, and Japanese; the build queue is German, Italian, Portuguese, Korean, Russian, Arabic, Dutch, Hindi, and Polish.
A1 through B2 at launch — four levels, from absolute beginner through upper intermediate. We're delaying C1 and C2 until we can source advanced vocabulary that lives outside frequency-based corpora (academic word lists, literary corpora). You set your level during onboarding and we tailor article complexity, deck size, and quiz difficulty accordingly. The monthly report will recommend a level-up when your data says you're ready.
It's an opt-in daily SMS or WhatsApp conversation. Each day at your chosen time, you receive a prompt tied to your current vocabulary. Reply in your target language, get gentle inline corrections, and continue for a few exchanges. Set it up during onboarding or in account settings.
Yes. Cancel from your account settings and your subscription ends at the next billing date. No questions asked, no friction.
Nothing bad happens. There are no streaks to break. The monthly report notes your engagement trend and may suggest adjusting difficulty or topic mix — that's it.