Privacy
Last updated: September 2026
Rhetorica transcribes conversations you choose to record. It uses the transcript to create scores, coaching tips, and Skill Reports. This page explains what we collect, who processes it, how long we keep it, and how to delete it. The app is in final testing. This is the policy planned for launch.
The short version
- Nothing is recorded until you start a session, and recording stops when you end it.
- Before every session, the app asks you to confirm that everyone present knows they’re being recorded and is happy about it.
- Audio is transcribed live, then discarded. We do not keep a recording. The transcript, analysis, and speech measurements remain until you delete the conversation.
- Transcripts and Skill Reports are yours, kept until you delete them.
- We never use your conversations to train AI models. We never sell them or use them for advertising.
- You can delete any conversation, or your whole account, at any time.
What the app collects
Your account. You sign in with Google or Apple, which gives us your email address and name. There are no passwords to collect, and we ask for nothing else: no phone number, no location, no contacts.
Your conversations. For each session you record: a transcript separated by speaker, the analysis built from it (scores, coaching tips, quoted moments), the conversation title, and any speaker names you type in yourself. The audio itself is transcribed as it streams and discarded when the session ends. We do not keep a recording. The app also keeps speech measurements for each speaker: pace, pauses, talk time, and filler words such as “um” and “uh.” Word timings are deleted after 90 days.
A consent record. Each time you start or resume a recording, the app stores the fact that you confirmed everyone present knew, the wording you agreed to, and when. It’s kept with the conversation and deleted with it.
What we deliberately don’t collect: the app contains no analytics, tracking, or advertising SDKs of any kind. It has no always-on or background listening mode. It builds no voiceprints. Speakers are told apart within a single conversation, but nothing identifies a voice across conversations.
How we use it
For one purpose: producing your transcripts, live coaching, and Skill Reports. We never use conversation data to train or improve AI models, never sell it, and never use it for advertising. Our AI providers are configured so they can’t train on it either. See the table below.
Recording other people
Rhetorica listens to the whole room, not just you. That’s why the app blocks recording behind a consent step every session: you affirm that everyone present knows you’re recording and is happy for you to. A visible recording indicator stays on throughout. Recording consent laws vary by country and by US state. Knowing and following the rules where you are is your responsibility, and the app is built for conversations where nobody needs to be surprised.
If someone recorded you
If someone recorded a conversation you joined, the transcript contains your words and a speaker label. It contains your name only if the account holder added it. We also keep measurements such as pace, pauses, and filler words for that conversation. We discard the audio after transcription and do not identify your voice across conversations. The service providers listed below process the audio or transcript. We do not use your words for training, sell them, or use them for advertising. Ask the person who recorded the conversation to delete it. This removes your words and speech measurements from our live systems. You can also email hello@rhetorica.app and we’ll help.
Who processes your data
Rhetorica doesn’t process audio on your phone. It streams to our servers and two AI providers. We name them because you shouldn’t have to guess:
| Service | What it receives | Its retention |
|---|---|---|
| Deepgram (speech-to-text) | Audio, streamed live during recording | Opted out of its model-improvement program: keeps content only as long as needed to process each request; technical request logs (no audio or text) up to 90 days |
| OpenAI (analysis) | Transcript text | Not used for training. Copies kept up to 30 days for abuse monitoring, then deleted unless the law requires otherwise |
| Railway (hosting) | Runs our servers and database | Follows the retention table below |
| Supabase, Google, Apple (sign-in) | Your sign-in identity. They do not receive conversation content. | Until you delete your account |
Processing happens in the United States under data-processing agreements with each provider. No one else receives your conversations.
How long we keep things
| Data | Kept for |
|---|---|
| Audio | Not stored as a recording. Audio is transcribed as it streams and discarded at the end of the session. Any temporary server buffer is gone within 24 hours. |
| Transcripts, analyses, Skill Reports, speaker names | Until you delete the conversation or your account |
| Speech measurements (per voice, per conversation) | Until you delete the conversation or your account |
| Word-by-word timings | 90 days, or until you delete the conversation, whichever comes first |
| Consent records | As long as their conversation exists |
| Account details (email, name) | Until you delete your account |
| Server working caches | Under 24 hours |
| Server logs (never containing conversation content) and encrypted backups | Up to 30 days |
Deleting
Delete any conversation or your whole account in the app. Deletion immediately removes the transcript, analysis, and consent record from our live systems. Account deletion also removes your identity from our sign-in provider. Backup copies expire within 30 days. Deletion is permanent. There is no undo or recovery window.
Security
Data moves over encrypted connections from your phone to our servers and on to the processors above. Audio is never written to permanent storage. It exists only for the moments needed to transcribe it. Your transcripts and reports are accessible only from your account.
Your rights
You can access, correct, export, or erase your data. You can manage conversations and your account in the app. For anything else, including export, email hello@rhetorica.app and we’ll respond within a month. If you’re in the UK or EU, these are your GDPR rights, and you also have the right to complain to your data-protection authority (in the UK, the ICO).
Children
Rhetorica is not directed at children under 13, and we don’t knowingly offer accounts to them.
Changes
When this policy changes, the date at the top changes with it, and material changes will be flagged in the app before they apply. We will never quietly weaken what’s promised here.
This website
This site is a static page. It sets no cookies, runs no analytics, and collects no personal information. There is nothing to opt out of, because nothing is collected.
Contact
Questions about privacy? Email hello@rhetorica.app.