This policy covers the Minutly AI web app and the Minutly AI apps for Android and iOS. It describes what the app does today, not what it may do later.
We do not use your recordings, transcripts or summaries to train any model of our own. We do not sell your personal data, and we do not share your content with anyone other than the processors listed below, except where the law requires it. We do not send you marketing email; privacy@getminutly.com and support@getminutly.com are mailboxes read by people.
On Android, Minutly asks for the phone state permission (READ_PHONE_STATE) for exactly one purpose: to pause your recording when a phone call starts and resume it when the call ends.
The app registers the system call-state listener, which reports one of three states: idle, ringing, or off-hook. A ringing phone on its own changes nothing. A move to off-hook (you answered, or you dialled out) pauses the recording. A return to idle resumes it. A call that rings and is then missed or declined is ignored entirely.
We do not record calls. We do not read your call log, your contacts, your SIM, your phone number, or device identifiers such as IMEI. The older Android callback hands our code a phone-number argument alongside the state; our code takes the state only and never looks at the number. The pause and resume state stays on your device and in the app's own screen — our server is told which audio segments arrived, not why there was a gap. The one exception is a recording that fails outright: then the failure message is saved on the meeting so the app can tell you what went wrong.
If you refuse the permission, recording still works. The app notes that call-pausing is unavailable and falls back on Android's audio-focus signal, which also catches most calls and voice-over-IP apps.
On iOS, no phone permission is requested at all. Call handling follows the standard system audio behaviour: when the system takes the microphone, recording pauses, and it resumes when the system gives it back.
To work at all, the app sends parts of your data to the processors below. Several of them are outside India, so your audio and transcript text can be processed outside India. We have checked each processor’s terms and state what each one commits to. Where a commitment started on a particular date, we say so, because it does not cover data sent before then.
Sarvam AI (Axonwise Private Limited, Bangalore): speech-to-text
Receives your recording audio in two ways. First, while you record in the browser (including in the iOS app), your microphone audio is streamed live to Sarvam in short chunks so that captions can be produced in real time, for up to two hours of a recording. Second, when a recording ends, the full audio file is sent to Sarvam for the final transcript: every mobile recording, and every browser recording up to two hours long. Audio recorded with the Android app is not streamed live; it is sent only as a file after you stop. No name, email or account identifier is sent with the audio. Sarvam’s policy is not to use customer content to train its models unless the customer opts in, and we have not. We do not control the region Sarvam processes in, so your audio may be processed outside India. Privacy policy: https://sarvam.ai/privacy-policy
OpenAI, L.L.C. (United States)
Receives more of your content than any other processor:
All requests go directly from our server to OpenAI under our own API account; they do not pass through any intermediary. OpenAI’s API terms do not permit it to train on this data, and we switched off every optional data-sharing setting in our account on 29 October 2026. Data sent to OpenAI is processed outside India. Privacy: https://openai.com/enterprise-privacy/
AssemblyAI, Inc. (United States): speech-to-text and speaker labels
Receives the full audio of browser recordings in four cases: to add speaker labels to recordings of 30 seconds or less, to transcribe recordings longer than two hours, as a fallback when Sarvam and Whisper both fail, and for uploads that could not be matched to a meeting. Mobile recordings are not sent to AssemblyAI. We have opted out of AssemblyAI’s model improvement program (31 October 2026), so it does not use our audio or transcripts to train its models, and it deletes them 30 days after processing. Processed outside India. Privacy policy: https://www.assemblyai.com/legal/privacy-policy
ElevenLabs, Inc. (United States): read-aloud
Receives the text of a summary, or of a transcript, only at the moment you press a Listen button, up to 5,000 characters, to synthesise the voice. If ElevenLabs is unavailable, your device’s own speech feature reads it instead and nothing is sent. Our account’s setting to use data for model training is off. Processed outside India. Privacy policy: https://elevenlabs.io/privacy-policy
Google LLC (United States)
Google handles sign-in. If you grant Drive access, a copy of each browser recording’s audio, and the transcript and task files you choose to sync, are saved to a folder in your own Google Drive. “Share as Google Doc” creates a document, including the full transcript, in your own Drive, private to you; you decide whether to share it from there. We do not send anything to Gmail: an email draft is built by our server and opened in your own mail app. Processed outside India. Privacy policy: https://policies.google.com/privacy
Emergent Labs Inc. (United States): hosting
Our application server runs on Emergent’s managed infrastructure, which uses Google Cloud and AWS, with processing in the United States, the EU and India. Emergent is bound by a Data Processing Agreement under which it does not use customer data to train models and requires the same of its sub-processors. Emergent is a hosting provider only; our requests to OpenAI and other processors do not pass through it. Privacy policy: https://app.emergent.sh/privacy-policy. DPA: https://app.emergent.sh/dpa
MongoDB, Inc.: database
Stores your account record, your meetings, transcripts, summaries, tasks and search embeddings, browser-recording audio, usage records, and any problem reports you send. Its region is a configuration setting we do not publish here, so your data may be stored outside India. Privacy policy: https://www.mongodb.com/legal/privacy/privacy-policy
Tigris Data, Inc.: object storage
Stores mobile-app recording audio only. Its region is a configuration setting we do not publish here, so your audio may be stored outside India. Privacy policy: https://www.tigrisdata.com/docs/legal/privacy-policy/
Resend: support email
When you send us a problem report from the app, the report is delivered to our support inbox through Resend: your email address, your note, the category, the meeting ID if you sent it from a meeting, a report ID, your account ID and an internal link for our support team, and, only if you tick the diagnostics box, your platform, app version, device, browser details, upload status and processing status. No transcript or summary is ever included. Nothing else is sent to Resend. Privacy policy: https://resend.com/legal/privacy-policy
We share data only with the processors listed above. Each one is bound by its own terms and privacy policy to protect the data it receives. For Sarvam, OpenAI, AssemblyAI, ElevenLabs and Emergent, the commitments and dates above are the ones that apply to your recordings, transcripts and summaries; under those settings and agreements, none of them is permitted to use that content to train its models. MongoDB, Tigris, Google and Resend store or deliver data and do not use it for AI training.
Your recordings, transcripts, summaries and tasks stay in your account until you delete the meeting or your account. When you delete your account we delete your user record, sign-in sessions, meetings, transcripts, summaries, tasks, stored audio (in our database and in Tigris), upload records and error reports, and we revoke our access to your Google account. If you choose the option during deletion, the ‘AI Meeting Notes’ folder in your Google Drive is moved to your Drive trash; otherwise the files you synced to Drive stay in your Drive under your control, as do any Google Docs you created. We keep problem reports you sent us, and usage records (minutes recorded, linked to your account ID) that we use for fair-use limits. We cannot delete copies that processors hold under their own retention rules; those are described above.
Recording is the one thing in Minutly that cannot happen without sending your data to the AI providers listed above. So we ask you once, on a screen of its own, before your first recording, and we record your answer against your account: which version of this policy you agreed to, when you agreed, and whether you have since turned it off.
What you are agreeing to is exactly what "Who Else Processes Your Data, and Where" describes: your microphone audio being streamed to Sarvam AI while you record in a browser so live captions can appear; your finished recording being sent to Sarvam AI (and, for short browser recordings, OpenAI and AssemblyAI) for transcription; and your transcript being sent to OpenAI to produce your summary, action items and follow-up email. No name, email or account identifier accompanies your audio. Some of those providers are outside India.
If you do not agree, nothing is created and nothing is sent — no meeting, no recording session, no audio. The rest of the app continues to work: you can still open, play, export and delete everything you have recorded before.
You can withdraw at any time in Settings, under "AI Processing". Withdrawing deletes nothing. It closes the door on new recordings and leaves every existing meeting, transcript, summary and audio file exactly where it is, fully readable and exportable. If a recording is already uploading when you withdraw, the audio already captured is kept on our servers rather than discarded, and it is not transcribed or summarised unless you turn AI processing back on. You can turn it back on whenever you like — changing your mind does not lock you out.
We keep a record of the agreement itself — the version, the date you agreed, the date you withdrew if you did, and whether the answer came from the web app or a mobile app. We keep the original agreement date even after a withdrawal, so there is an honest history of what was agreed and when. This record is deleted with your account.
If what we send to our AI providers materially changes, the agreement you gave stops counting and the screen appears again, explaining that the terms changed, before your next recording. That is how the promise in "Changes to This Policy" below is actually enforced rather than merely stated.
Two honest notes. This agreement covers AI processing only; it is not the basis on which we hold your account or your existing recordings. And the shared demo account is shared — an agreement given while signed in to the demo is recorded against the demo account, not against you, and applies to anyone else using it.
Minutly AI's use and transfer of information received from Google APIs to any other app will adhere to the Google API Services User Data Policy, including the Limited Use requirements.
Minutly AI requests only non-sensitive Google OAuth scopes: basic sign-in and drive.file. That one Drive scope covers both the "AI Meeting Notes" folder and the Google Docs you export, and it can only ever touch files Minutly itself created — it gives us no access to anything else in your Drive. We request no sensitive and no restricted scopes at all. Basic sign-in is requested at sign-in; drive.file is requested on demand, the first time you use a Drive-backed feature.
Basic sign-in (openid, userinfo.email, userinfo.profile)
Used solely to authenticate the user and display their name, email, and profile picture in the app. We never use this data for marketing.
Google Drive (drive.file) — non-sensitive, and the only feature scope we request
Used only to access files that Minutly AI itself creates within the user's Drive. We cannot access any other files in the user's Drive. We use this one scope for both things: saving audio recordings and transcript and task files into a folder our app creates ("AI Meeting Notes"), and creating and formatting the Google Doc a user exports a meeting to. The drive.file scope is the most restrictive Drive scope and Google APIs prevent our app from reading any pre-existing files the user did not create through Minutly AI.
No separate Google Docs scope is requested
Exporting a meeting to a Google Doc runs entirely under drive.file: Google's Docs API accepts that scope for creating and formatting a document the app itself created. Earlier versions also requested the sensitive documents scope. That is gone. We never read or modify any document a user did not create through Minutly AI, and there is no separate Docs permission to grant — one Drive connection covers it.
No mail scope is requested
Minutly can prepare a follow-up email draft from the meeting summary and open it in the user's own email app; Minutly does not send email and has no access to the user's mailbox. Earlier versions requested a Gmail send scope to deliver that email on the user's behalf. That is gone: no mail scope is requested, no mail API is called, and nothing about a user's mail is read, searched or stored.
In compliance with Google's Limited Use requirements, Minutly AI:
drive.file scope) and a copy remains on Minutly servers with the meeting until you delete it. Enabling Drive sync does not remove our copy.Traffic to our servers uses HTTPS, and the Android app is configured to refuse plaintext connections. Every request for your content is filtered by your account, so one account cannot read or change another's. Error stack traces are deliberately unreadable through any signed-in screen. Session cookies expire after 30 days, and account deletion revokes our Google tokens with Google.
You have the right to:
Minutly AI is not intended for use by children under the age of 13. We do not knowingly collect personal information from children under 13. If we learn that we have collected such information, we will delete it promptly.
We may update this Privacy Policy. The date at the top of this page reflects when it was last revised. Before any material change takes effect, we will ask you to review it and agree again in the app.
If you have questions about this Privacy Policy or how we handle your Google user data, please contact us:
We aim to respond to all privacy inquiries within 5 business days.