Kaiwaflow is an AI app for real Japanese conversation practice. It takes the words you've already studied and turns them into spoken conversations, so you actually say them out loud.
You've banked thousands of words. Anki decks, WaniKani levels, vocab you can recognise in a heartbeat. But the moment you have to say something, nothing comes. Paste in your own word lists and notes, and Sensei builds real conversations around them: your people, your places, your grammar. The vocabulary you've drilled for months finally comes out of your mouth.
Recognition isn't production. You can read the word, pick it out of a sentence, ace the review, and still freeze when it's your turn to talk. Input apps and spaced repetition (SRS) got you a huge passive vocabulary; nothing turns that passive vocabulary into active speaking.
Kaiwaflow is built for intermediate learners. If you're already past the basics, the textbook lesson plan is the wrong shape. You don't need streaks, leaderboards, or another quiz on past-tense conjugation. You need a partner who remembers your context, meets you mid-conversation, models the natural form when you stumble, and knows who Mac is the next time you mention him, speaking the dialect you are working on, the vocabulary from your last review.
That's the only thing Kaiwaflow tries to be.
Hold space, speak, release. Sensei replies in your ears, not just your eyes. Optional fullscreen mode for hands-free practice with subtitles you can toggle off. High Fidelity Voice available with Pro.
browser speech-to-text · Whisper coming soonPaste a word list, drop in a CSV, or dump freeform notes. Kaiwaflow extracts vocabulary and context in one pass, and remembers your context between sessions. Sensei knows about the Hokkaido trip, the dog you lost, and the grammar from last Tuesday, then weaves it into what she says next.
paste or CSV · LLM-driven extractionSensei never interrupts to correct. Instead, she models the natural form back in her reply. For the curious, a small chip below each message lets you see exactly what was tacitly fixed.
a different kind of language appPaste a vocab list, upload a CSV, or drop in your raw study notes. No accounts to connect. Your words go straight in.
An AI pass pulls vocab + context from whatever you pasted. Skim, fix, skip what isn't vocab. The same view handles tidy CSVs and messy prose.
Sensei opens with a greeting that mentions someone or somewhere from your notes. Speak or type, whichever fits the moment.
Save anything Sensei says as an Anki card. Your library lives alongside the chat, browsable and exportable. One card at a time.
Kaiwaflow is the output half of your practice. It's the place you finally use what immersion, tutoring, and Anki put in. Keep all of them.
Sensei won't interrupt to red-pen you. She models the natural form back in her reply, and a small chip shows what she quietly fixed.
Pro's High Fidelity Voice sounds like a person, not a GPS. When the monthly cap is hit, your browser's voice takes over and Sensei keeps talking.
We won't sell you a timeline. What you get is reps speaking your vocabulary out loud. It's the thing most apps skip.
Pro is one flat $19 per month, or $180 a year (save 21%). No usage meters, no overage fees, no per-message charges. You always know what you'll pay.
We cap High Fidelity Voice at ~12 hours of voice per month so a handful of power users don't break the math for everyone. Hit the cap and your browser's built-in voice takes over. Sensei keeps talking, just at lower fidelity until the cap resets on the first of the month.
Generous daily limits run quietly in the background to prevent abuse and keep flat pricing sustainable. Normal use never touches them.
Try Kaiwaflow's core experience with no card required.
Everything, including every language we ship next. Founding members lock today's price for life. It only goes up from here. Auto-applied trial; cancel before day 15 and you pay nothing.
no card up front · cancel anytime
Kaiwaflow is a web app for spoken Japanese conversation practice. You import vocabulary you have already studied — from Anki, WaniKani, JPDB, Migaku, plain text, or CSV — and Sensei, the AI partner, builds real conversations around it so the words you can recognise finally come out of your mouth.
Kaiwaflow has no streaks, XP, or levels and is not a beginner curriculum. It is built for learners past the basics who already have vocabulary but freeze when it is time to speak — it gives you speaking reps using your own words, not gamified lessons.
Anki builds passive recognition; Kaiwaflow turns that passive vocabulary active by having you speak it. You import your decks and Sensei weaves your words into spoken conversation. It is the output half of your practice, not a replacement for Anki — keep using both.
Yes — speaking is the core. You hold space to speak, Sensei replies with voice, and Pro adds High Fidelity Japanese voice (Whisper transcription is coming soon). There is an optional hands-free fullscreen mode with subtitles you can toggle off.
Kaiwaflow does not interrupt to correct you. Sensei models the natural form back in her reply, and a small chip below the message shows exactly what was quietly fixed — flow over correction.
There is a free tier — free chat, browser voice, CSV import, save up to 100 vocab words, and 30 chat turns per day, with no card required. Pro is a flat $19 per month, or $180 a year, with no usage meters and no overage fees, and the 15-day Pro trial needs no card up front.
Yes. From Settings → Billing, a monthly subscriber can switch to yearly in one click — Stripe prorates the change automatically, crediting the unused part of the current month against the yearly price. You can switch back to monthly anytime from the Stripe billing portal.
You can cancel anytime from Settings → Billing. When you cancel you keep Pro until the end of the period you have already paid for, then your account moves to Free. We do not issue refunds for partial periods. Plan switches (monthly to yearly) are prorated automatically by Stripe rather than refunded.
Anki, JPDB, WaniKani, and Migaku exports, plain text, or CSV. An AI pass extracts the vocabulary and context in one step, then you skim and fix before you start talking.
Yes — that is exactly what Kaiwaflow does. Import your Anki deck or WaniKani vocabulary and those exact words are scheduled into conversation: each one resurfaces naturally in spoken practice when it is due, and your recall is verified when you actually produce it. No other app closes that loop.
Retrieve it in conversation. Recognition and production are different skills — more flashcard reviews build the first, not the second. Kaiwaflow surfaces your spaced-repetition-due words in speech, so you produce the vocabulary you have studied instead of just recognising it.
Yes. Sensei remembers your context across sessions — your vocabulary, the people and topics you mention, and which words you still struggle with. Unlike ambient chatbot memory, this recall is scheduled and verified: due words are deliberately woven back into conversation, and the app confirms when you produce them.
We're letting people in a few at a time. Leave your email and we'll send an invite as spots open.