How it works.
Spaced repetition you already know, wrapped around content that isn't generic.
Why generic flashcards fail.
You can finish a 1,500-card frequency deck and still freeze when the moment arrives. The cards were correct. They were just about nothing in particular.
The problem with frequency-list decks isn't accuracy. It's that the words show up disconnected from any moment in which you'd actually use them. You learn them at the kitchen table and can't retrieve them in the room where you need them.
Spaced repetition, generative content.
We don't reinvent spaced repetition. The scheduler is the same one Anki users already trust. What's different is upstream of the scheduler: where the cards come from.
You describe a conversation you have coming up. The system generates a small deck — usually three to seven cards — about that moment. The doctor you're seeing on Thursday. The class meeting on Tuesday evening. The phone call to the bank you've been putting off.
Cards reviewed the night before a conversation are easier to recall the next morning, because the moment is what is doing the encoding — not just the spaced-repetition curve.
The language-agnostic engine.
The engine is built around a translation pair and a scenario, not around any particular language. We launch with English → Finnish. The next pairs — EN→SV, FI→SV — are already working in private builds and will surface as the audience grows.
If you live somewhere else and are facing similar conversations, write in. The list of pairs we open next is roughly the list of people who asked.
What we don't do.
No streaks. No gems. No anthropomorphic mascot to sulk at you for skipping a day. No "league" of strangers. No push notifications guilting you into seven more minutes.
You moved to another country. You can decide when to study. Our job is to make the moments when you do study feel useful, not engineered to keep you in the app.