Privacy policy
Still Fluent (the iPhone app, sometimes called Language Keeper during development, and this website) is made by Just A Guy Software LLC ("we", "us"). This policy explains what we collect, why, who else sees it, and how to delete it. We've tried to write it in plain English.
Short version. We collect what the app needs to run your practice and nothing for advertising. We don't sell your data and there are no ads, trackers or analytics in the app or on this site. To hold a conversation with you, the app sends what you say to AI providers (Anthropic, and the voice provider you pick). Nothing about your practice is shown to friends unless you turn it on for that friend. You can delete your account in the app at any time.
What we collect
Your account
- Email address, used to sign you in (by emailing you a sign-in link) and to contact you about the beta. If you use Sign in with Apple, we receive the email Apple gives us, which may be a private relay address, and an identifier Apple assigns for this app. We never see your Apple ID password.
- If you continue with Google, Google tells us your email address, name and profile photo link, plus an identifier Google assigns. Our sign-in provider stores these with your account; the app uses only the email. We never see your Google password.
- Username and display name, if you set them. Your username is how friends find you.
Your learning profile
- The language you're practicing (Mandarin or Japanese), simplified or traditional characters, and whether readings (pinyin or furigana) are shown.
- The topics you're interested in and your daily reminder time.
- Your self-rated levels for listening, reading, speaking and writing (on the ILR scale). We keep each rating you give so you can see how they change.
Your practice
- Sessions: when you practiced, for how long, and which piece you read.
- Saved words and their review schedule.
- Mistakes and corrections from your conversation feedback (your error journal), including the original phrase, the correction and how often it recurs.
- Voice usage: how many minutes of cloud voice you used, with which provider, and our estimated cost, so we can apply monthly limits.
- Beta voice ratings: if you rate a voice after a conversation, your scores and which provider it was.
Conversations
When you talk with the AI partner, the app produces a text transcript of the conversation. The transcript is sent to Anthropic to generate replies and your end-of-session feedback (see below). We keep the feedback (the mistakes, corrections and new words above). We don't currently keep full conversation transcripts on our servers; if that changes, we'll update this policy first.
To help you practice what you're relearning, each conversation also includes up to eight practice items: saved words that are due for review and mistakes from your error journal that keep coming back. The partner is asked to work them in naturally, and they go to the same provider as the rest of the conversation.
Friends
If you use the friends feature: friend requests, who your friends are, invite links you create, and the sharing level you've chosen for each friend.
What we don't collect
- No advertising identifiers, no location, no contacts, no photos.
- No analytics or tracking SDKs in the app, and no analytics, cookies or trackers on this website. If you sign in to practice on the website, your browser keeps your sign-in in its local storage on that device until you sign out.
- We don't store recordings of your voice.
Your voice and speech
The Talk step has two kinds of voice, and you choose which to use:
- Cloud voices (OpenAI, Google Gemini or xAI Grok). While a conversation is live, audio from your microphone streams directly from your iPhone to the provider you picked, along with the piece you're discussing, your speaking level and a few practice items (see below), so it can listen and reply out loud. The provider also returns a text transcript. Our server only hands out a short-lived access key for each session and records the minutes used; the audio doesn't pass through or get stored on our servers.
- Apple on-device voice. Your speech is turned into text by Apple's speech recognition, on the iPhone itself where the device supports it (on older devices Apple may process it on its servers under Apple's privacy policy). The text goes to Anthropic for a reply, which your iPhone reads aloud.
The app asks for microphone and speech recognition permission before Talk is used. You can turn either off in iPhone Settings, and you can always type instead.
Who we share it with
We use these service providers to run the app. Each gets only what it needs for its job, and none of them may use your data for advertising on our behalf.
| Provider | What for | What they receive |
|---|---|---|
| Supabase | Database, sign-in, and the server functions the app calls | Everything listed under "What we collect." |
| Resend | Delivering sign-in emails | Your email address and the sign-in link we send you |
| Anthropic (Claude) | Conversation replies, end-of-session feedback, word definitions | Conversation transcripts, the piece you're reading, your language and level, and up to eight practice items |
| OpenAI, Google (Gemini), xAI (Grok) | Live voice conversation, only the one you pick, only while you're talking | Your voice audio as you speak, the piece you're discussing, your speaking level, and up to eight practice items |
| Google (sign-in) | Continue with Google, only if you choose it | That you're signing in to Still Fluent. Google sends us the details listed under "Your account." |
| Apple | Sign in with Apple, on-device speech, TestFlight distribution | Whatever you share with Apple directly (for example TestFlight feedback or crash reports, if you choose to send them) |
| Vercel | Hosting this website, including the web version of the app | Standard request logs (such as IP address and pages requested) when you visit the site. When you practice on the website, your conversation requests pass through Vercel to our backend; Vercel doesn't store them |
We use these companies' paid developer services, not their consumer apps. How they handle what they receive, including how long they keep it for safety and abuse monitoring, is set by their own terms and privacy policies, linked above. We will not opt in to letting them use your data to train their models.
Other than that, we share information only if the law requires it, to protect someone's safety, or as part of a sale or transfer of the business (in which case this policy would still apply to your data). We don't sell or rent personal information, and we don't share it for cross-context behavioral advertising.
Friends and what they can see
Friends see nothing about your practice unless you choose to share with them. Anyone who knows your username can send you a friend request, and friends see your username and display name. Once you accept, you pick a sharing level for that friend: nothing (the default), streak only, streak and minutes, or full activity (sessions, language and progress). You can change it, per friend, whenever you like. Friends never see your saved words, mistakes or conversations.
How long we keep it
We keep your data while you have an account. Invite links expire after seven days. When you delete your account, your account and practice data are deleted from our database right away. Copies may remain in our hosting provider's encrypted backups for a short period (up to 30 days) before they're overwritten.
Deleting your account and your other choices
- Delete everything: in the app, go to Settings → Delete account. This removes your account and all practice data and can't be undone.
- Sign in with Apple: you can also stop using your Apple ID with the app in iPhone Settings → your name → Sign-In & Security → Sign in with Apple.
- Continue with Google: you can remove the app's access at myaccount.google.com/connections.
- See, correct or export your data: most of it is visible and editable in the app. For a copy of everything, or anything else, email us at hello@justaguysoftware.com. We'll respond within 30 days.
Depending on where you live (for example the EU, UK or California), you may have legal rights to access, correct, delete or port your data, or to object to some processing. We'll honor those requests from anyone, wherever they live. We process your data to provide the service you asked for (performing our agreement with you) and, for things like security and abuse prevention, our legitimate interests.
Where your data is
We're based in the United States and our providers may process data in the United States and other countries. By using the app you understand your information will be handled there.
Security
Data travels over encrypted connections. Our database only lets each account read its own rows; friends' data reaches you only through a check of the sharing level they set. API keys for the AI providers stay on our server and never ship in the app. No system is perfectly secure, and we'll tell you promptly if a breach affects your data.
Children
Still Fluent is for adults. It isn't directed at children under 13 (or under 16 where that's the local age of digital consent), and we don't knowingly collect their data. If you think a child has an account, email us and we'll delete it.
Changes
If we change this policy, we'll update the date at the top. If a change is significant, such as sending your data to a new kind of provider, we'll tell you in the app or by email before it takes effect.
Contact
Just A Guy Software LLC
hello@justaguysoftware.com