Privacy Policy

Who we are

BraveVoice is a speech confidence app for children, operated by Lone Orca Ventures and powered by an AI coaching character named Terry. For privacy questions, contact us at: privacy@bravevoice.me

What we collect

We collect only what is necessary to provide the service:

• Child's first name and age range: used to personalise Terry's responses. We never collect a child's full name.
• Interests and topics mentioned: used to make conversations relevant. Stored locally and optionally synced to your account.
• Session history and exercise progress: used to track growth over time. Stored locally and optionally synced to your account.
• AI-generated notes about what your child has shared: Terry remembers a small number of things your child mentions in conversation (interests, names of friends, things they're working through) so future sessions feel consistent rather than like meeting a stranger each time. These are generated automatically from session transcripts, capped in number, and used only to personalise Terry's coaching, never shown to your child directly, never used for any purpose beyond that session-to-session continuity.
• Speaking-confidence profile (optional), if you tell us your child stutters, experiences speaking anxiety, or both, and if you further identify a specific pattern (for example, going quiet in specific settings, or avoiding certain situations), we store that classification to shape how Terry coaches your child. You control what's shared here and can change or clear it at any time from your dashboard.
• Avoidance moments you log (optional), if you use the dashboard feature to note that your child avoided a speaking situation (e.g. a phone call, a class discussion), that entry is stored so Terry can gently and privately take real-world context into account. Never shown to your child, never referenced by name.
• Parent email address: only if you create an account. Used for login and to sync data across devices.
• Birthday month (optional), used so Terry can acknowledge your child's birthday month. We collect month only, never day or year.
• Payment information: only if you subscribe. Collected and processed directly by Stripe; we never see or store your full card number.

How we use data

• To personalise Terry's coaching for your child
• To track progress and display it in the parent dashboard
• To sync data across your devices if you have an account

We do not use your child's data for advertising, profiling, or any purpose beyond delivering the service.

How data is stored

All session data is stored locally on your device using browser storage. If you create an account, data is also synced to our secure cloud database (Supabase), hosted in the United States. Data in transit is encrypted via HTTPS.

Children's privacy (COPPA)

BraveVoice is used by children as young as 3 through teenagers, but every child's account is set up, owned, and overseen by a parent or guardian regardless of the child's age. For any child under 13, we are designed with COPPA principles in mind and are actively working toward full compliance verification with the Children's Online Privacy Protection Act (COPPA). We require verifiable parental consent before a child may use the app. Specifically:

• A parent or guardian must complete the consent screen before first use
• We do not knowingly collect personal information from children without parental consent
• Parents may review, update, or delete their child's data at any time by contacting us
• We do not condition a child's participation on providing more information than is necessary

AI conversations

Terry's responses are generated by OpenAI's API (the GPT-4o mini model). If OpenAI is temporarily unavailable, we automatically fall back to Groq's API so a session isn't interrupted; in that case the conversation content for that exchange is processed by Groq instead. Terry's voice is generated with OpenAI's text-to-speech API. On iPhone and iPad, speech is transcribed with Groq's hosted Whisper API. In desktop browsers such as Chrome, speech is transcribed by the browser's built-in speech recognition, which is processed by Google. OpenAI does not use API data to train its models, and it keeps API request logs for up to 30 days for abuse monitoring before deleting them (OpenAI: your data). Groq processes data as our service provider under its Services Agreement and Data Processing Addendum. We send these providers only what is needed to run the conversation: your child's first name, age range, interests and the session's messages, never your email address or payment details.

Terry is instructed to flag any messages involving self-harm, danger, or possible abuse disclosure. An independent, non-AI keyword check also runs on every message as a backstop, so a flag doesn't depend solely on the AI's own judgment. These flags are stored securely in the cloud (for users with an account) and surfaced to parents in the dashboard. For messages the system identifies as involving self-harm, the crisis resources Terry provides (the 988 Suicide & Crisis Lifeline and the Crisis Text Line) are fixed, pre-written text, not AI-generated, to make sure that specific response is always accurate. For possible abuse disclosure, Terry is guided to encourage the child to tell a trusted adult, and in some cases also mentions the Childhelp National Child Abuse Hotline (call or text 1-800-422-4453), for some phrasings this is the same kind of fixed, pre-written text as the self-harm resources above; for others it is part of Terry's in-character, AI-generated response.

Terry is an AI, and our safety protocol

Terry is an AI character, not a person. Every session starts with an on-screen reminder that Terry is a pretend dinosaur made by a computer, and Terry answers honestly if your child asks whether it is real. If the app stays open for three hours, it shows a break reminder that repeats this. Like any AI companion, BraveVoice may not be suitable for every child. You know your child best.

Suicide and self-harm. When a message suggests self-harm or suicidal thoughts, Terry does not discuss methods or continue the topic. It replies with fixed, pre-written text that refers your child to the 988 Suicide & Crisis Lifeline (call or text 988) and the Crisis Text Line (text HOME to 741741). The rest of the session is held in a gentle, no-challenge mode, the parent is emailed right away, and Terry checks in kindly at the start of the next session. Detection uses both the AI's judgment and an independent keyword check on every message.

Possible abuse, and requests for pictures. Possible abuse disclosures are handled the same way, with the Childhelp hotline (1-800-422-4453). If anyone asks your child for pictures, Terry gives a fixed reply telling them they never have to send any and to tell a grown-up, and for older children points to Take It Down. Terry never produces sexual content.

Drugs, alcohol and vaping. Terry answers questions like these with care and never gives instructions. It suggests talking with a trusted grown-up and adds a gentle "worth a chat" note to your dashboard and the session email.

Dangerous dares. If your child asks about a dare that can stop someone breathing (such as the "blackout" or "choking" challenge), Terry gives a fixed, pre-written reply saying it is dangerous and to tell a grown-up or call 911 if someone is hurt, and you are emailed right away.

Third-party services

• OpenAI: Terry's responses (primary) and Terry's voice (text-to-speech). OpenAI Privacy Policy
• Groq: Backup for Terry's responses, and speech transcription on iPhone/iPad (Groq's Whisper API). Groq Privacy Policy
• Google: Browser-based speech recognition on desktop (Chrome). Google Privacy Policy
• Supabase: Cloud database for account sync. Supabase Privacy Policy
• Resend: Transactional email delivery (magic link sign-in, session and progress emails). Resend Privacy Policy
• Vercel: Hosting for the BraveVoice website and its server functions. Vercel Privacy Policy
• Stripe: Payment processing for parent subscriptions. Stripe receives billing details and payment method directly; we never see or store your full card number. Stripe Privacy Policy

Data retention

We keep your child's data for as long as your account remains active, so that progress and history are available across sessions. If you remove a child's profile from your dashboard, that child's full record, session history, memory notes, avoidance log, alerts, and profile details, is deleted from our database immediately. So that the free trial (5 sessions per family) can't be reset, we keep only a count of how many sessions a removed child had used, with no session content. We do not currently have an automatic deletion schedule for inactive accounts; if you'd like your data removed at any time, contact us using the methods below and we will delete it within 30 days.

Data deletion

To delete all data for your child:

• Self-service: If you have more than one child profile, use "Remove child profile" in your dashboard to delete that child's full record immediately.
• Local data: Clear your browser's site data for bravevoice.me
• Account data: Email us at privacy@bravevoice.me and we will delete all associated records within 30 days, this is the only path available if you have only one child profile, since the dashboard requires at least one profile to remain.

Changes to this policy

We will post updates to this page and update the "last updated" date above. For material changes that affect how we collect or use children's data, we will notify parents by email before the changes take effect and obtain fresh consent where required by law.

Contact

Questions about this policy or your child's data: privacy@bravevoice.me