Privacy policy

What we collect and why

• Your email and account identifier for sign-in; your timezone and reminder time to schedule reminders; and records of your age confirmation and recording consent.
• Typed ideas, voice recordings, transcripts, titles, AI summaries, key points, questions, scores and their explanations so you can capture and review ideas.
• If you join the beta waitlist without an account, just the email address you enter and which form you used, so we can invite you. Email hello@buzzdraft.co to be removed.
• Reminder and sharing records. If you enable notifications, we store push subscription endpoints, delivery keys and browser information to send reminders. An idea title can appear on your lock screen.
• Product events named capture_started, capture_saved, summary_ready, reminder_sent, reminder_opened, idea_kept, idea_shared, paywall_viewed and plus_waitlist_joined, linked to your account identifier. Properties describe capture mode, connection state, timings, score, delivery counts, opening source and link reuse. They do not include idea text, titles, summaries, transcripts, emails or share tokens. The upgraded event is reserved for future payments and is not sent yet. Events are stored in our database; forwarding to PostHog is active only when configured.
• Technical error information in server logs to troubleshoot failures. When configured, Sentry receives error reports and a small sample of performance traces. We scrub request bodies, cookies, headers, content fields, error messages and share tokens before reports leave the app. Session replay is disabled.

Recording and consent

We ask for consent before your first recording and show a visible indicator while recording. You must get everyone's consent before recording a group. Permission to use your microphone does not give you permission to record other people.

By default, we keep raw audio only until it has been successfully transcribed and the transcript saved, then automatically delete it. Failed transcriptions keep the audio so a retry can work, until transcription succeeds or you delete the idea or account. If audio cleanup itself fails, the recording can remain until cleanup succeeds on a retry or you delete it with the idea or account.

Recordings are stored in a private bucket. Playback links, when used, are signed rather than public.

AI processing and service providers

The services below process data to run BuzzDraft when configured. Development versions can use simulated AI results instead.

• Supabase: database, authentication and private audio storage.
• OpenAI: voice recordings sent for transcription.
• Anthropic: typed ideas or transcripts sent for AI summaries.
• Inngest: background processing and reminder jobs, when enabled.
• Vercel: planned production hosting.
• PostHog: product events and sanitized pageviews, only when configured.
• Sentry: scrubbed error reports and sampled performance traces, only when configured.
• Resend: planned sign-in email delivery through Supabase Auth.

OpenAI's and Anthropic's API data terms say inputs and outputs are not used to train their models by default. BuzzDraft does not train models on your ideas. Provider account settings still need confirmation before launch; these defaults do not mean providers retain no data.

• OpenAI API data controls
• Anthropic commercial data policy

When you share an idea

Anyone with a share link can see the card: its title, summary, key points, score, score explanation and creation date. It does not include the transcript, raw typed text, recording or account details.

We automatically mask detected emails, phone numbers, links and @handles in the card. This can miss details and does not remove people's names. Check the content before sharing.

Links expire 30 days after creation. You can choose Stop sharing at any time. Deleting the idea also stops its links. Someone may already have saved a copy, screenshot or preview; stopping the link cannot erase those copies.

Storage on your device

Sign-in uses cookies. The app saves captures, including recordings, on your device before syncing, so you can capture offline. It also keeps preferences and cached pages for offline use. Unsynced captures remain on that device until synced or discarded. Use a device you trust and clear its site data when you no longer want local copies.

When browser analytics is configured, PostHog identifies signed-in users by account identifier only and records page changes with dynamic identifiers removed from paths. It respects Do Not Track. Automatic click and text capture and session recording are disabled. Its browser identity is reset on sign-out and account deletion.

Keeping and deleting your data

Saved ideas and transcripts stay in your account until you delete them. Archiving an idea does not delete it.

• Choose Delete on an idea and confirm to remove the idea, its transcript, summary, remaining audio, reminder and share links.
• Choose Delete account in Settings and complete the confirmation to remove your account, ideas, transcripts, summaries, audio, reminders, shares, push subscriptions and account-linked database events. When PostHog management credentials are configured, we also request deletion of your PostHog person record and associated events. This external cleanup is best effort; a failure does not stop deletion of your account in BuzzDraft.

If deletion fails, the app asks you to retry. These controls remove data from the app's active storage; this draft does not promise immediate erasure from provider backups or logs. Provider retention periods and deletion arrangements still need review before launch.

Changes and privacy questions

We will update this page and its date when our data practices change. For questions or help with your data, contact: