Pick your lake
Pick A1-1, A1-2, A2-1 or whichever fits. You see the full map and choose where you stand today — we don't guess for you, and there's no "where do I start?" paralysis.
Every CEFR sublevel — A1-1, A2-3, B1-2 — is a finite, mappable knowledge lake. You see everything inside and drain each one at your own pace, in five formats. No more blind spots, no more pretending you're moving forward.
No card. German, French, Spanish and English — the whole map, open.
Built for brains that need to see the whole thing — especially if yours runs ADHD or neurodivergent.

Try it now — no signup
A small taste of A1-1 in any of the four languages. Hit "Talk now" and start a conversation.
Alemán · A1-1 · with Lukas
2-minute demo. Practice greetings and basic phrases.
The conscious self-directed learner roughly knows which CEFR sublevel they're at. What they don't have is a visible map of what's inside each lake, a way to drain them in the format their brain asks for today, and the certainty that they're not skipping anything.
Duolingo decides what you practice today. You don't see what lakes exist or which ones it's skipping — only the next step in the corridor.
ChatGPT improvises with off-curriculum vocabulary. It generates waves in any direction — and forgets the lesson when you close the tab.
Babbel and Memrise drop you in a fixed lake, in a single format, with the same tone for a dev, a doctor and a teenager.
Without a visible map and without an editable history, reviewing means walking blindfolded. For an ADHD or neurodivergent brain that dark zone isn't discomfort, it's paralysis: you don't start because you don't know where, and the streak in whatever app you're using finishes sinking you.
You pick a lake, generate a lesson, leave a trace. The friction is where it should be — in learning, not in navigating.
Your pace. Your lake. Your way.
Pick A1-1, A1-2, A2-1 or whichever fits. You see the full map and choose where you stand today — we don't guess for you, and there's no "where do I start?" paralysis.
You pick a format (summary, quiz, exercises, flashcards or conversation) and a style. The AI builds the lesson on top of that lake's real vocabulary.
You mark done, rate low/good/high and schedule a reminder. Your history is real — not a counter of consecutive days.

No pressure. Or keep scrolling for more details
Duolingo walks you down a hidden route. ChatGPT generates ripples without a map. Babbel drops you in a single lake, in a single format. Aprandr does something different: it shows you the full map of CEFR lakes and the AI helps you drain each one in the format your brain asks for today.
The AI is the engine. As it gets better, we open new formats so every lake can be walked through the way your brain absorbs it best.
Every feature answers a concrete pain of the self-directed learner. Starting with the biggest one: seeing the map.
Every sublevel — A1-1, A2-3, B1-2 — is a knowledge lake mapped by a human expert. You see the full set of vocabulary, phrases and rules before you start. The AI doesn't invent the syllabus: it operates inside that lake.
It's not Duolingo picking the route for you, nor ChatGPT being asked "teach me German".
Summary, quiz, exercises, flashcards and conversation. Same lake, five ways to walk through it depending on how you learn best today. Built for brains that get bored with one format — you pivot when one tires you out, not when the app lets you.
It's not a linear course that comes in a single pre-cooked format.
Talk live with an AI tutor — from the browser or by phone. It follows a fixed loop: announces the lake item, makes you produce it, closes and moves to the next one. It corrects your structure when you get it wrong.
It doesn't grade pronunciation. Only structure and correct use of the active lake's vocabulary.
Ask for the same lesson "like to a dev", "like to a 10-year-old" or in academic tone. The wrapper changes, the lake doesn't.
Not translating the same PDF into another font.
Mark done, rate low/good/high and schedule reminders. Your history is real — which lake you touched, when, with what quality.
Not a streak that punishes you for getting sick on a weekend.
Filter by status, quality or reminder. The navigation adapts: you go from the lake map into the lesson tabs without losing context.
Not a flat dashboard with 200 mixed-up items.
German, French, Spanish and English — with the full lake map. More languages are coming, but only once we have the curated catalog of every sublevel ready.
We don't promise 40 languages just to have 40 empty maps.

Five principles — the first one is the structural bet: showing you the entire map. The other four are the mechanics of how you drain each lake.
Each CEFR sublevel is a knowledge lake — vocabulary, phrases and rules curated by a human expert. You see it whole before you start. No surprises, no "hidden routes", no algorithm deciding for you. Especially important if your brain runs ADHD or neurodivergent — seeing the whole thing is relief, not decoration.
You set the horizon (1 wk / 1 mo / 3m / 6m / custom). It's not an opaque SM-2 algorithm — the aware learner knows what they retain best.
Quiz forces a choice before revealing the answer. Inline inputs make you produce before validating. Flashcards show the front first. You produce, you don't just read.
The AI biases generation toward the active lake's lines — one step above your current level (i+1), not random. The style adjusts lexical density without breaking focus.
Five ways to drain the same lake: quiz for discrimination, exercises for guided production, conversation for contextual comprehension, flashcards for lexical speed, summary for grammar. You pick the muscle of the day.
It's the metaphor we use to think about every CEFR sublevel. A lake is finite and mappable: it has a fixed set of vocabulary, phrases and rules curated by a human expert. You see it whole before you start — nothing is hidden behind an algorithm. The promise: no more invisible lakes, no more pretending you're moving forward. "Sublevel" is the technical name; "lake" is how we talk about the product.
Because Duolingo decides the route for you — you don't see what lakes exist, which ones you've already touched or which ones it's skipping. You move through a linear corridor without knowing what's around you. Aprandr does the opposite: it hands you the full map from A1 to B1 on day one and you decide which lake to drain today.
ChatGPT generates random ripples — it doesn't know which vocabulary belongs to A1-2 vs A2-1, so it improvises off-curriculum. Here every lesson is built from the curated catalog of the lake you picked. The content stays inside, and your history is saved — it doesn't evaporate when you close the tab.
It doesn't guess. You pick the lake. If you're not sure, start at A1-1 and move up when it feels easy. There's no mandatory placement test — that friction doesn't serve the self-directed adult.
Yes — it's designed for that, without being only for that. Most product decisions answer pains the ADHD brain lives daily: you see the full map of lakes (the dark zone of "I don't know what I'm missing" disappears), you pick the lake — not what to study inside it (less decision fatigue), you generate the lesson when hyperfocus hits and leave it half-done if it crashes (no big deal), and you have five formats to pivot to when one gets boring. No streaks, no XP, no shame. If you're not ADHD it works too — it's just that an ADHD brain feels seen using it.
Because the streak turns into an obligation. When it breaks — and it breaks — the intrinsic motivation you had to learn the language gets replaced by shame, and shame kills the habit. The ADHD brain pays double: a streak broken on Friday switches off Monday. Aprandr doesn't compete with your motivation using an external counter. A bad day breaks nothing; the map is still there when you come back.
Mark it as low quality and generate another. It's on demand, you're not tied to a single version. Your rating stays in your history and the "Quality: high" filter becomes your personal curation over time.
No. Lesson generation depends on cloud AI. Already-generated lessons can be reopened as long as you have a connection.
German, French, Spanish and English — with the full lake map from A1 to B1. We're curating the catalogs for what comes next before we ship them. We don't want to launch empty languages just to inflate the list.
Yes, to start. There will be a paid plan when we stabilize generation limits. What you generate now stays with you.
Yes. Every lesson you generate is tied to your account, with its status (pending / done), quality (low / good / high), and reminder. You can come back to it whenever.
Today we go up to B1 — the B2 and C1 lakes are still being curated. If you're higher than that, it's not for you yet. When we ship those levels you'll see the entire map, same as with A1-B1.
You pick a CEFR lake, a format, and watch what the AI puts together in under 10 seconds. If it doesn't work for you, rate it low and generate another — the map doesn't move.
No pressure. No streaks. Your first lake whenever you want.