Privacy Policy
Last updated: 2026-09-28
Provided in English and Japanese. If the two differ, the Japanese text prevails.
This Privacy Policy explains what personal information the operator of Clerfy (the "Service", "we") collects when you use it, why, who processes it, and the choices you have. The Service handles sensitive documents, so we keep this policy concrete.
1. Information we collect
- Account: your email address, the name and language your sign-in provider (Google or Apple) shares, and your language and notification settings.
- People you add: the name and relationship (spouse, child, parent, other) of family members you manage, and their language.
- Facts about your situation: the answers you give in the Service — status of residence, dates such as the end of your period of stay, employer category, school type, and similar — and the procedures you create.
- Documents: images and PDFs you upload (for example a residence card, passport, tax certificate or employment certificate), the fields the Service reads from them (names, dates, issuer, fiscal year), and the checks it ran.
- Questions you ask the assistant and the answers it gave, together with the rules it used.
- Feedback you send us through the feedback form: what you wrote, the kind of request, an optional 1–5 answer and the procedure it concerns.
- Technical and usage data: device and browser type, event names such as "document uploaded" (without the content), error logs, and cookies described in section 8.
- How the site and the app are used: the pages you visit and the buttons you use, recorded by our own analytics server (self-hosted open-source software on the same infrastructure as the Service, not a third-party service). Before anything is recorded, identifiers are removed from page addresses and only a few campaign and setting parameters are kept. When you are signed in, this is linked to a random pseudonym, never to your name, email address or account number. Your approximate location (country, region, city) is derived from your IP address; the IP address itself is not stored.
2. What we deliberately do not collect
The Individual Number (マイナンバー). The Service is designed never to store it: documents that can show the 12-digit number are not read until you confirm that your copy hides it, the text of PDFs is masked before any AI model sees it, and the model is instructed not to transcribe it. If you notice the number in a document you uploaded, delete the document and upload a copy without it.
We do not collect information for advertising, we do not build profiles for third parties, and we do not sell personal information.
3. How we use information
- to find the procedure that matches your situation, build your checklist from the published rules, and keep your dates and next step;
- to read the documents you upload and compare them with what you entered and with the published requirements;
- to answer your questions using the rules applied to your procedure;
- to send you the notifications you enable (deadlines, document freshness, rule changes);
- to keep the Service secure, prevent abuse, and fix errors;
- to understand how the Service is used, in aggregate and without document content, and to measure the cost and quality of AI features.
Rule decisions — which requirements apply, deadlines, dates — are made by deterministic software over published rules, never by an AI model. AI is used to read documents, classify them, explain and translate.
4. Who processes it and where
We use the following processors under contracts that limit them to processing on our instructions:
- Supabase (database, authentication, file storage) — servers in Tokyo, Japan.
- Vercel (web hosting and the application API) — functions run in Tokyo, Japan.
- Anthropic (AI reading of documents and answers to your questions) — processed in the United States, under terms that prohibit using your data to train models. Only the masked text or the confirmed copy of a document is sent, together with the facts needed to explain it.
- Google and Apple (sign-in) — they tell us who signed in; we do not receive your password.
- Modal (batch workers that draft rule updates from official pages) — official sources only, no personal information.
Because some processing happens outside Japan, your information may be transferred to countries with different data-protection laws. We rely on contracts with the processors above to protect it (Act on the Protection of Personal Information, Art. 28).
5. Who can see your information
Only you and the people you explicitly grant access to (for example a spouse, or a professional with a time-limited grant). There is no employer or school dashboard; an organisation that pays for the Service in the future will not see anything you have not chosen to share.
We disclose information to authorities only where the law requires it, and we will tell you where we are allowed to.
6. How we protect it
- Encryption in transit and at rest.
- Documents live in private storage and are opened only through short-lived links; every access to a document is recorded.
- Access control at the database row level, so one account cannot read another's rows.
- Production logs do not contain document contents or extracted values.
- Secrets and credentials are stored encrypted and scanned for exposure.
7. Retention and deletion
- Procedures and documents: kept until you delete them in the Service. Deleting a document also deletes the stored file.
- Assistant logs and AI usage records: kept for up to 90 days for quality and cost analysis; they never contain document contents.
- Account: to delete your account, contact us at the address below. We delete the account, the people it manages that no other account can access, and their procedures and documents, within 30 days, except where the law requires us to keep a record.
- Backups are encrypted and cycle out within the same period.
8. Cookies
The Service uses a sign-in session cookie and cookies that remember your language (chosen with the language switch or by opening a page in a language, such as /vi) and appearance choice; signing out removes the preference cookies. Right after you sign in, a cookie that lasts at most five minutes notes whether you just created your account, and is deleted once read. Usage analytics (section 1) do not use cookies. It uses no advertising or cross-site tracking cookies. Usage analytics record event names only, without document contents or personal identifiers beyond what is needed to count unique users.
9. Children
Accounts are for people aged 18 or over. Information about a child is entered only by a parent or guardian managing the child's procedure, and is treated as described in this policy.
10. Your rights
Under the Act on the Protection of Personal Information you may ask us to disclose, correct, add to, delete, or stop using your personal information, and to tell you how we obtained it. You can correct or delete most of it yourself in the Service; for anything else, contact us at the address below. We will respond without undue delay, after confirming your identity.
11. Changes
We may update this policy. The date at the top shows the current version; material changes are announced inside the Service. This policy is provided in English and Japanese; if the two differ, the Japanese text prevails.
12. Contact
Personal information handling business operator: Clerfy — support@clerfy.app
The operator's address and the name of the person responsible are disclosed without delay on request to this address.