YOUR REAL LIFE STAYS YOURS.

Rehearsal content. Setup answers, messages, AI replies, and coaching reports are transmitted to our servers and our AI processor to generate the rehearsal and report and check the request against our safety and product boundaries. On the web, they are also stored in your browser so a rehearsal can survive navigation and refreshes. The iOS app keeps the active transcript in memory and does not restore it after the app closes.

Practice records. We may store a device or account identifier, scenario and opponent identifiers, practice objective, score, subscores, derived behavior signals, attempts, and outcomes. We do not put raw transcripts or recordings on public leaderboards.

Voice rehearsals. With your permission, BigTalk records your microphone while you answer. The iOS app requires supported on-device speech recognition; it falls back to text rather than sending your microphone audio to BigTalk or Apple for transcription. On the web, speech recognition is supplied by your browser and may be processed under the browser provider's terms. The resulting text is processed like any typed rehearsal turn.

Runbacks. Voice clips remain in browser memory or temporary app files. BigTalk does not upload or store those recording files on its servers. Character audio is selected by default. Including your own reply requires a separate choice. Your device renders the selected audio into a captioned clip for you to review. The clip leaves your device only when you deliberately share, save, or download it. Copies you share are then governed by the destination service and the people who receive them.

Accounts. If you sign in, we process your email address, authentication records, and the practice history linked to your account. Anonymous use relies on a random identifier stored in an HTTP-only cookie.

Callouts. A private Callout stores the sender's first name, opponent, two practice scores, random-link hash, and opened, started, and completed timestamps. The bearer link expires after seven days. The raw secret is never stored in the database.

Purchases and live services. We may receive purchase, subscription, booking, and entitlement records. Payment card details are collected directly by the payment provider and are not available to BigTalk.

Private Reserve referrals. We store a random referral code, the anonymous or account identifiers for inviter and invitee, and attribution and qualification timestamps. A referral qualifies after the invitee completes a scored rehearsal. The referral record does not contain the rehearsal transcript, audio, score, or coaching report.

Depending on the configured service, BigTalk uses Anthropic or OpenAI for scenario setup, roleplay, coaching, and safety checks, directly or through Vercel AI Gateway. ElevenLabs, or xAI when that voice service is selected, synthesizes the fictional counterpart's speech from text. Generated speech can include details from your rehearsal. These providers process requests under their own terms and our service relationship with them. Provider retention and training controls differ; do not assume that a private rehearsal means zero provider retention. Do not enter information you do not want processed by an AI service.

BigTalk is a practice tool, not therapy, medical care, legal or investment advice, crisis support, or a prediction of what another person will do.

We use data to provide and secure rehearsals, generate feedback, save progress, process purchases, fulfill bookings, prevent abuse, operate private Callouts, support users, and understand whether practice improves outcomes.

We do not use private rehearsal content for advertising or expose it to employers, teammates, Callout recipients, or the public unless you deliberately share an artifact. Score cards omit the transcript. Runbacks contain only the exchange you review and choose to share.

We use service providers including Anthropic and OpenAI for language generation and safety checks, ElevenLabs and xAI for counterpart speech, Supabase for database and authentication services, Vercel for hosting and AI routing, Stripe for web payments, and Google when you choose Google sign-in. Each provider processes data under its own terms and our service relationship with it.

Web rehearsal data remains in browser storage until you clear it, replace it with a new rehearsal, or delete your data. The iOS app does not persist its active transcript between launches. Unshared voice files are held only in browser memory or temporary app storage and are cleared when you start over or the session ends; the operating system may also clear temporary files. Files you download or share remain where you put them. Expired Callout records are periodically deleted. Account and practice records remain while your account is active unless a shorter operational period applies.

You can delete account-linked and anonymous practice data from the in-productSettings page. Limited transaction, fraud-prevention, tax, or legal records may be retained when required, with account identifiers removed where practical.

BigTalk is not directed to children under 13. Do not use the service if you are under 13. Users under the age of majority should use BigTalk with permission from a parent or guardian where local law requires it.

Private Reserve is limited to adults age 18 and older. It is not directed to or intended for minors.

You may use the core product without creating an account, clear local data, delete an account, decline optional sharing, or contact us about access, correction, deletion, or privacy questions.

Privacy contact: support@thebigtalk.ai