Integritetspolicy

Senast uppdaterad: 28 september 2026

1. What We Collect

When you use Proudfrog, we collect the following information:

• Email address — used for account authentication and to send you emails about your recordings and your account.
• Audio recordings — recorded with the Proudfrog iOS app or macOS desktop app, or uploaded as files in the web dashboard, for transcription. While you record, audio is saved on your device in segments and then uploaded for processing.
• Transcripts and metadata — generated from your recordings, including speaker labels, timestamps, and AI-generated summaries and notes.
• Voiceprints — a numeric description of a speaker's voice, used to recognise the people you name and to suggest names for recurring voices in your own meetings (see Section 5).
• Meeting context — notes, photos, images, and files you choose to attach to a meeting.
• Calendar details — if you allow calendar access, the title, time, and attendees (names and email addresses) of the calendar event that overlaps a recording are sent with it, to suggest a meeting title and speaker names.
• Purchases — a record of the credit you buy and use. Card payments are handled by Stripe; we never see or store your card details.
• How you found us — when you create an account, we save on it where your first visit came from: an ad click ID (such as Google's gclid), campaign (UTM) parameters, the host name of the site that linked to us, and the page you landed on. It comes from the link you followed or, if you accepted cookies, from the pf_attr cookie (see Section 7). We use it only to see which of our marketing brings in users, based on our legitimate interest in measuring our own marketing. It is kept as long as your account.

2. iOS App — On-Device Data

The Proudfrog iOS app accesses the following on your device, each only after you grant permission:

• Microphone — to record meetings. The iOS app records the sound in the room through the microphone; it does not capture audio from other apps or calls on your phone.
• Camera and photo library — optional, to attach photos (for example of a whiteboard) to a meeting. Only the images you pick are uploaded.
• Calendar — optional and read-only, to match a recording with its calendar event and attendees. The app never creates, changes, or joins events.
• Speech recognition — optional, for dictating notes. Dictation uses Apple's speech recognition service.
• Notifications — optional alerts, for example when your recording is paused due to an interruption (e.g. a phone call).
• Local storage — recordings, speaker reference clips, and app preferences are stored on-device until they have been uploaded. The app checks available disk space to prevent recording failures.

The app does not access your contacts or location. It does not contain any tracking or advertising SDKs.

3. macOS Desktop App — On-Device Data

The Proudfrog macOS desktop app accesses the following on your device, each only after you grant permission:

• Microphone — to record your voice during meetings.
• Screen & System Audio Recording — used solely to capture system audio (e.g. the other participants in a video call). Proudfrog does not capture any video or screen content — only the audio output from your system. macOS requires this permission for any app that accesses system audio.
• Calendar — optional and read-only, to match a recording with its calendar event and attendees. The app never creates, changes, or joins events.
• Local storage — recordings, context notes, file attachments, and app preferences are stored on-device until they have been uploaded.

The app does not capture screenshots, video, or any visual content from your screen. It does not access your files or contacts beyond what is listed above. It does not contain any tracking or advertising SDKs.

4. How We Use Your Data

• To transcribe your audio recordings with speaker separation.
• To generate meeting notes, key terms, summaries, and action items, and to answer your questions about your meetings, using AI language models.
• To recognise speakers you have named, and suggest names for recurring voices, in your own meetings (see Section 5).
• To send you emails about your recordings and account.
• To provide the web dashboard for reviewing and editing transcripts, and the MCP server if you connect an AI assistant.
• To handle credit purchases and usage-based billing.
• To see which of our ads and campaigns bring in new users (see Section 7).

Your audio recordings and transcript text are sent to the third-party services listed in Section 6 for processing. The app requests your explicit consent before any data is shared with these services. We do not use your content to train AI models.

5. Röstigenkänning

När du namnger en talare, eller godkänner ett föreslaget namn, sparar Proudfrog ett röstavtryck för den personen: en uppsättning siffror som beskriver rösten, inte en inspelning av den. Senare möten där samma röst känns igen med hög säkerhet läggs till i avtrycket. För att hitta röster som återkommer i dina möten sparar vi också ett röstavtryck för talare du ännu inte har namngett, och det används bara för att föreslå namn och gruppera samma röst i ditt eget konto. Ett sådant avtryck raderas när talaren får ett namn (då tar den namngivna personens avtryck över), när mötet eller kontot raderas och i alla fall efter högst 180 dagar. Låter du iOS-appen lära sig din egen röst när du börjar, sparas även den korta inspelningen i ditt konto.

Röstavtryck räknas som biometriska uppgifter enligt GDPR. Vi använder dem bara för att känna igen och föreslå röster i dina egna möten. De lagras i EU tillsammans med resten av ditt konto, syns bara i ditt konto och delas aldrig med andra användare (inte heller när du delar ett möte), används aldrig för att träna modeller och säljs aldrig.

Du kan när som helst glömma en röst: öppna personen under Personer och välj Glöm rösten, så raderas röstavtrycket medan namnet och historiken ligger kvar. Du kan också glömma hela personen från ⋯-menyn på samma sida, och då raderas röstavtrycket med den. Raderar du ditt konto raderas alla dina röstavtryck. Du ansvarar för att berätta för dem du spelar in att du använder Proudfrog, och för att ha deras samtycke där lagen kräver det.

6. Third-Party Services (Subprocessors)

We use the following third-party services to run Proudfrog. The list may change over time; we will update this policy when it does. All providers are required to protect your personal data to the same or an equivalent level as described in this policy.

Transcription and AI

• AssemblyAI (United States) — speech-to-text transcription and speaker separation. Audio recordings are sent to AssemblyAI for processing. See their privacy policy.
• Anthropic (Claude) (United States) — AI language models for summaries, meeting notes, cleanup, key terms, and conversational search. Transcript text is sent to Anthropic for processing. See their privacy policy.

Infrastructure & other services

• Hetzner Online — servers in Helsinki, Finland (EU) that run our API, background processing, and MCP server, and object storage for audio files.
• Supabase — user authentication and database (PostgreSQL), hosted in the EU.
• Stripe — payment processing for credit purchases.
• Resend — email delivery.
• WorkOS — sign-in for the MCP server, when you connect an AI assistant such as Claude, ChatGPT, or Cursor.
• Google and Apple — optional sign-in with your Google or Apple account.
• Sentry — crash reporting and error monitoring in our apps and the web dashboard, including short session replays of the dashboard in which all text, input and media are masked before they leave your browser. No audio or transcript content is sent to Sentry.
• Vercel — hosting for the web dashboard and this website.

7. Cookies and Analytics on Our Websites

proudfrog.com and the web dashboard at app.proudfrog.com use a few cookies. Cookies that are not needed for the sites to work are only set if you accept them in the cookie banner. You can change your choice at any time under “Cookie settings” at the bottom of proudfrog.com; the choice applies to both sites.

Always on

• pf_consent — our own cookie that remembers whether you accepted or declined cookies, so we don't ask on every visit. Kept for 12 months.
• NEXT_LOCALE — remembers the language you picked on proudfrog.com. Kept for 12 months.
• Sign-in cookies on app.proudfrog.com (Supabase Auth) — keep you signed in to the web dashboard.
• Umami — visitor statistics that we run ourselves (page views, referring site, device type). Umami uses no cookies and does not follow you across other sites.

Only if you accept

• Google Ads (Google Ireland Ltd.) — conversion measurement: when you arrive through one of our ads and then download an app or create an account, Google Ads counts it, so we know which ads are worth paying for. Cookies _gcl_au and _gcl_aw (and related _gcl_* cookies) on proudfrog.com, kept for up to 90 days. We use Google Consent Mode: until you accept, Google's tag sets and reads no cookies and strips the ad click ID from what it sends. Google then only receives cookieless signals that a page was viewed or a conversion happened, which it uses for aggregated, modelled statistics. Google may process data in the United States. See Google's privacy information.
• pf_attr — our own cookie on proudfrog.com, kept for 90 days. It remembers where your first visit came from (ad click ID, UTM campaign parameters, referring site, landing page and time), so that app.proudfrog.com knows which marketing brought you if you create an account. If you decline, nothing is stored: the same parameters are only passed along in the link when you click through to the app.
• Microsoft Clarity (Microsoft Corporation) — heatmaps and session recordings that show how people use proudfrog.com and the web dashboard (clicks, scrolling, where people get stuck), so we can fix what doesn't work. Clarity is not loaded at all before you accept. Cookies _clck (1 year) and _clsk (1 day). Microsoft may process data in the United States. See Microsoft's privacy statement.

If you withdraw your consent, we delete pf_attr and the Google Ads and Clarity cookies from your browser.

8. Data Storage

Your account data, transcripts, and audio files are stored in the European Union: on Hetzner servers and object storage in Helsinki, Finland, and in Supabase's EU infrastructure. To be transcribed and analysed, audio and transcript text are transferred to AssemblyAI and Anthropic in the United States (see Section 6). All data is encrypted in transit (HTTPS/TLS).

9. Data Retention

Audio recordings, transcripts, and associated data are kept for as long as your account is active, so you can play back and review your meetings. Audio is not deleted after processing — it stays in your account until you delete the recording. You can delete individual recordings and transcripts from the web dashboard at any time.

If you delete your account, all associated transcripts, speaker data, and stored audio files will be permanently removed within 30 days.

10. Your Rights

You have the right to:

• Access the personal data we hold about you.
• Request deletion of your data.
• Export your transcripts.
• Withdraw consent at any time by deleting your account.
• Change or withdraw your cookie consent at any time under “Cookie settings” at the bottom of proudfrog.com.
• Object to our use of data about how you found us (Section 1); contact us and we will remove it from your account.

11. Children's Privacy

Proudfrog is not directed at children under the age of 13. We do not knowingly collect personal information from children. If you believe a child has provided us with personal data, please contact us at privacy@proudfrog.com and we will delete it promptly.

12. Security

All data is transmitted over HTTPS (TLS). Sign-in (email, Google, or Apple) is handled by Supabase Auth with JWT tokens; sign-in to the MCP server uses OAuth via WorkOS. We do not store passwords directly.

13. Contact

Proudfrog is operated by Upnorth AI AB (org.nr 559495-0882), Stockholm, Sweden, which is responsible for your personal data. For privacy-related questions or data requests, contact us at privacy@proudfrog.com.