About Tomodekki
The best Japanese learners alive today use eight to twelve tools at once. That stack is genuinely powerful. It is also hostile to anyone who just wants to learn the language.
A serious learner's setup usually looks something like this: Anki for vocabulary, a browser dictionary for reading, a separate app for kanji, another for grammar, a video player with dual subtitles, a frequency list, a textbook, and a Discord full of people explaining how to wire them together. Every one of those tools is good. The seams between them are where people quit.
Tomodekki is an attempt to make the integrated stack the default product, without throwing away the depth that the obsessives correctly insist on.
Four things we believe
Capture beats curation
A word saved in two seconds is worth ten perfect cards that never got made. The most valuable moment in learning a language is the one where you notice something you don't know — on a sign, in a subtitle, in a conversation you half-followed. Almost every app makes you wait until you're home and at a keyboard. By then the moment is gone. Tomodekki takes voice, camera and text from wherever you are and builds the card afterwards.
AI is the middleware, not the headline
You should see a beautiful card, not a chatbot. The model's job is to work out the reading, choose a natural example sentence using words you already know, and get out of the way. If a feature is only impressive because AI made it, it isn't a feature.
Spaced repetition is a means, not the product
Reviews should feel like checking a feed, not doing tax returns. The scheduler is the most consequential thing in the app and it should also be the least visible — you answer honestly, and it works out when to ask again.
Your life is the curriculum
There is a built-in path that takes you from your first kana outward, ordered so you never meet a kanji before its parts. But what you actually read, watch and overhear should flow back into the app, not the other way around.
Where it's up to
The app is built and running on real devices: the scheduler, the capture surfaces, the card library, pronunciation and pitch-accent work, the curriculum. What it hasn't had is people who aren't us using it. That's what early access is for, and it's why the signup form asks where you're up to — a cohort of people who already read Japanese would tell us almost nothing about whether a beginner can get started.
If you want to be part of that, put your email in. If you'd rather ask something first, get in touch.