Legal
Privacy Policy
Last updated 2026-08-09
Why this matters here
Roji (“the Service”) is for teachers — language coaches and others — to keep a record of how their students are doing and decide what to teach next. Alongside the teacher's own details, we hold information about that teacher's students. So we're explicit about what we collect, what we use it for, and what we never do with it.
What we collect
- The teacher's email address, for signing in
- Workspace name and the subject taught
- Students' display names, goals, assignments, submissions, learning records and teacher comments
- A student's email address, if they've been invited to sign in themselves
- Uploaded files — handouts, photos, audio
- Usage records: how much AI was used, and error logs
- What you enter in the contact form: topic, email address, name and message
If you sign in with Google
When you sign in with Google, all we receive is your email address and display name. We use them to identify you and to contact you about your account.
We never access any other Google service — not Gmail, Drive, Contacts or Calendar. The only scopes we request are email address and basic profile.
We never use anything received from your Google account for advertising, analytics, or sale to third parties. You can revoke the connection at any time from your Google account security settings.
What we deliberately don't collect
We don't collect information that teaching doesn't need. There are no fields for a postal address, phone number, date of birth or gender. A student's display name only has to be something the teacher recognises — it doesn't have to be their legal name.
What we use it for
- Running the Service: storing and showing records, delivering assignments, receiving submissions
- Generating AI drafts: analysis, assignment suggestions, study plans
- Answering your enquiries — replying to the address you gave us
- Investigating faults and improving the Service
- Billing
About the AI
We use AI (Anthropic's Claude) to draft assignments and analyse progress. Doing that sends the necessary part of a student's learning record to the AI provider.
- We never let your data be used to train AI models. Our agreement with the provider excludes training.
- Nothing AI generates is shown to a student until the teacher has reviewed and confirmed it.
- We send learning content only. Email addresses and other contact details are never sent.
Sharing and sub-processors
We don't share your information with third parties without your consent, except where the law requires it. To run the Service we use the following sub-processors for storage and processing.
- Supabase — database and file storage. Data is stored in the Tokyo region
- Vercel — running the application
- Anthropic — AI text generation
- Stripe — card payments. We never hold card numbers
Some of these are outside Japan, so your data may be processed outside Japan.
How we keep it separate
Data is separated by organisation, and the database itself enforces that one teacher can't see another's (row-level access control). When a student signs in, they see their own published assignments and what the teacher has confirmed — nothing else. Drafts, and AI output the teacher hasn't confirmed, are never shown.
How long we keep it
We keep records until the teacher deletes them, or until a workspace is closed and deletion is requested. Archiving a student keeps their records but hides them from the student and takes them off your bill. To have data deleted, contact us.
Access, correction and deletion
To see, correct, restrict or delete your own information, contact us at the address below. We'll confirm your identity and act on it.
Contact
Not set yetUse the contact form at /legal/contact.
Changes to this policy
We revise this policy when we need to. We'll announce any significant change inside the Service.