Privacy

MentorStar is built for young learners (ages 5–7) and school partners. We design for data minimization, pseudonymous classroom join, and evidence-grade learning events — not open surveillance or free-form chat logs.

What we collect in product

• Session and mission events (start/end, attempts, hints, mastery changes).
• Pseudonymous learner ids (e.g. via class code join) — not child email accounts.
• Optional parent transfer verification (Yes / Not yet).
• Device/class identifiers needed for sync and classroom ops.

What we do not collect by design (POC)

• Open free-text chat between children and an unrestricted LLM.
• Diagnostic labels or peer ranking in the child UI.
• Continuous video of the child as a default product path.

How data is used

Events power mastery progress, teacher live cues, parent Proof Cards, pilot exports, and ops dashboards. Aggregated summaries may use AI (e.g. Gemini) to draft pilot narratives from metrics — not to chat with children.

Storage & security

Devices store events offline-first. Cloud sync may dual-write to Firestore and optional Postgres. Firestore security rules block direct client writes; server Admin SDK and authenticated API tokens perform ingestion.

Parents & schools

Schools run pilots under their policies and consent processes. Parents see trust-first Proof Cards and transfer suggestions. For pilot data requests or deletion, contact your school pilot lead and pilots@mentorstar.app.

Website

This marketing site may use standard hosting logs. Admin tools are not for public indexing.

This page is a product privacy overview for the POC. Formal DPA / school contracts will supersede for paid pilots.