Privacy policy
What this app is
Daftar is a personal voice journal. You speak; we transcribe, summarise, and help you search and reflect on what you said. Egyptian Arabic (Masri) and mixed Arabic/English are first-class.
What we collect
- Account email (sign-in codes)
- Voice recordings you create, and text derived from them (transcripts, titles, summaries, mood, people, topics, commitments)
- Optional photos you attach to an entry
- Usage meters needed for fair-use limits (audio seconds and Ask count per calendar month)
- Subscription status from our payment provider (RevenueCat / the App Store)
- App preferences (language, timezone, audio retention)
- Local device reminders you enable (scheduled on your phone; we store the remind time)
We do not sell your journal. We do not use your entries to train our own models. Some processors’ default terms allow them to use API inputs to improve their services (see Voyage below).
Where it is stored
Primary app database, auth, and file storage run on Supabase in the United Kingdom
(London, eu-west-2). The United Kingdom is covered by an EU adequacy
decision for personal data transfers; we do not claim “EU-only” hosting.
How we process it
To run the product we send relevant data to:
| Processor | Role | Region notes |
|---|---|---|
| Supabase | Database, auth, file storage, edge functions | United Kingdom (London) |
| ElevenLabs | Speech-to-text | Per their DPA / terms (audio leaves our storage to be transcribed) |
| Anthropic | Summaries, weekly/monthly reviews, Ask | Per their DPA / terms; API traffic is not used to train their models under their current API terms |
| Voyage | Embeddings for search | Per Voyage AI’s terms: unless a customer opts out, they may use API inputs (e.g. text we send to embed an entry or search query) to train and improve their models. Daftar does not use Voyage’s training opt-out, so that default applies. We only send text needed for search embeddings (typically title, summary, and transcript), not audio files. |
| RevenueCat | Subscription entitlement sync | Purchase events only; not journal content |
| Apple | App Store handles payment | Payment data stays with Apple; sign-in codes are emailed by us |
Recordings and derived text are stored encrypted at rest. We do not claim end-to-end encryption: the server must read entries to transcribe, summarise, and search.
Journal content (including mood) may be treated as special-category data under GDPR. We ask for explicit consent in the app before the first recording.
Access and money
Reading your past entries, Timeline search, export, and deletion work without a paid plan. Recording/processing, reviews, and Ask require an active trial or subscription. The server enforces this; the paywall is only UX. Subscriptions are billed through the App Store (and later Google Play if offered).
Your controls
- Export: download a zip of entries, transcripts, and any audio still stored
- Delete audio, keep text: per entry or all at once
- Delete photo on an entry (when you attached one)
- Audio retention: optionally auto-delete recordings older than 30 / 90 / 365 days
- Delete account: permanently removes your auth user, database rows (cascade), and storage objects under your prefixes
Safety
If processing flags possible self-harm or crisis language, the app may show a calm card with public helpline links for your region. No alert is sent to anyone else, and the journal is not blocked.
Retention
- Account and journal text: until you delete the account (or individual entries)
- Audio: until you delete it, retention expires, or the account is deleted
- Photos attached to entries: until you remove them or delete the entry/account
- Exports: short-lived signed URLs (minutes); objects under your export prefix until deleted with the account
Contact
Changes
We will update this policy when processors, regions, or controls for Daftar change. Published versions show an effective date.