Last updated: September 1, 2026
Readersmith is a voice-first reading tutor for young children. It is built by a parent, for families, and this policy is written the same way: plainly. The short version — there are no accounts, no ads, no data sales, and no third-party analytics. Your child's words are used to run the tutoring conversation, and our servers keep nothing from a session unless a grown-up has given consent in the app.
On your device only: your child's profile (the name and emoji you pick), their reading progress, and app settings. This stays on the device. We never see it.
Voice audio: when a reading session is running, the microphone streams your child's voice to our server, which relays it to Google's Gemini API — that conversation is the tutor. We do not save recordings of your child's voice, and our production servers keep the words of the conversation out of their logs unless a grown-up has consented (see below). Google processes the audio as our service provider under its Gemini API terms.
On our server: an anonymous device identifier (a random token — no name, email, or phone number is ever collected), which invite code a device used and when, per-device usage counters (sessions, minutes, and spending caps that keep the service safe to run), and technical diagnostics about sessions — timings, audio quality signals, and event counts, not words.
Optional, off by default: nothing more is collected unless a grown-up grants consent in the app's grown-up settings. With that consent, the app sends usage data — which activities were tried and how they went, how long things took, the reading language and the app's language, the app version and device type, crash reports, and an identifier for each reader that is not their name — and our server keeps each session's transcript (the words the tutor said and heard) and your reader's first name. The app shows exactly this list before you consent. Everything collected under that consent is deleted after 90 days.
Readersmith is made for children who cannot yet read, and it is set up and controlled by a parent or guardian: access requires a family invite code that a grown-up enters, and all settings live behind grown-up surfaces. We collect no personal information about your child — no name, no age, no contact details — unless a grown-up consents to analytics, in which case the reader's first name and the session transcript are kept for 90 days, as described above. The one sensitive thing the app touches, your child's voice, is used only in the moment to hold the tutoring conversation and is never stored.
The operational records described above (usage counters, invite redemptions, technical diagnostics) are kept while we operate the service. Anything collected under a grown-up's consent — usage data, crash reports, session transcripts, the reader's first name — is deleted after 90 days. Deleting a reader, or deleting everything, in the app's grown-up settings erases that reader's (or the device's) records on our server as well as on the device. You can also email us to have everything associated with your device deleted.
Questions or deletion requests: aran.donohue@gmail.com.
If this policy changes, the date above changes with it, and material changes will be called out in the app's grown-up settings.