Your conversations never leave your Mac.
This page documents exactly what Remark does with your data. It is written for IT teams, compliance officers, and anyone who needs to verify our privacy claims before approving Remark for use in their organisation.
What processing happens where
Every component of Remark's transcription, formatting, and storage pipeline runs locally on your Mac. Your voice recordings, transcriptions, and meeting data never leave your device. Remark downloads speech recognition models on first use. After that, all processing is local.
| Process | Technology | Runs on |
|---|---|---|
| Speech-to-text transcription | On-device speech recognition model | Apple Silicon (local) |
| Speaker diarization | On-device speaker identification model | Apple Silicon (local) |
| Text formatting & cleanup | On-device language model | Apple Silicon (local) |
| Audio capture (meetings) | macOS system audio | Local audio pipeline |
| Audio capture (dictation) | Microphone input | Local audio pipeline |
| Search & summarisation | On-device inference | Apple Silicon (local) |
Remark does not use OpenAI, Anthropic, Google, or any other cloud AI provider. The models are downloaded on first use and then run entirely on Apple Silicon hardware.
What data is stored
All data created by Remark is stored as standard files on your Mac's local filesystem. There is no database server, no cloud sync service, and no proprietary storage format.
Files created by Remark
- Audio recordings (WAV/M4A) — saved to a folder you choose
- Transcripts (plain text / Markdown) — saved alongside recordings
- Meeting summaries and action items — embedded in transcript files
- Speaker voice profiles — stored in Remark's application support directory
- Application preferences — stored in macOS standard defaults
You can move, copy, back up, or delete these files at any time using Finder or any file manager. Remark does not prevent you from accessing or modifying your own data.
License key storage
If you purchase Remark Pro, your license key is stored in Remark's local app preferences on your Mac. License validation is performed entirely offline on your Mac — no network request is made.
What leaves the Mac
Your voice recordings, transcriptions, and meeting data never leave your Mac.
Remark makes zero network requests during transcription, dictation, recording, formatting, search, or export. The application does not contain analytics, telemetry, tracking pixels, usage monitoring, or any form of data collection.
There are exactly three scenarios where Remark communicates over the network:
1. Initial model download
On first use, Remark downloads speech recognition models required for transcription. This is a one-time download. After the models are cached locally, all transcription and audio processing happens entirely on your Mac with no further network requests.
2. Update checks (Sparkle)
Remark uses Sparkle, an open-source macOS update framework, to check for new versions. This sends a standard HTTP request to our public GitHub repository to check the appcast XML file. No user data, device identifiers, or usage information is included in this request. You can disable automatic update checks in Settings.
3. Crash reports (optional, off by default)
If you choose to enable crash reports in Settings, a report is sent when the app crashes. Crash reports contain exactly four pieces of information:
- Error type (e.g., "EXC_BAD_ACCESS")
- Stack trace (function call sequence leading to the crash)
- macOS version (e.g., "14.4")
- Remark version (e.g., "1.2.0")
Nothing else. No audio. No transcripts. No meeting content. No device identifiers. No usage patterns.
You can also report issues via email. Remark opens your default email client with a pre-composed message. You see everything that will be sent before you hit Send. Nothing is transmitted without your explicit action.
We will never add analytics, telemetry, tracking, or usage monitoring. If it is not listed on this page, it is not being sent. This commitment is architectural, not just policy — the code simply does not contain the infrastructure to collect or transmit user data.
Independent verification
You do not need to take our word for any of the above. Remark's privacy claims can be verified independently:
- Use Little Snitch, LuLu, or any network monitor to verify that Remark makes no outbound connections during transcription, dictation, or formatting
- Inspect the application bundle to confirm that no analytics SDKs or tracking libraries are included
- Run Remark in airplane mode to confirm it works without any network connectivity
- Examine your filesystem to see exactly what files Remark creates and where
If your IT or compliance team requires additional information, contact us and we will provide whatever technical documentation you need.