Privacy Model
Understand what stays local and what can leave your device in MachinesFluent.
MachinesFluent is private-first because it gives you local speech recognition options. Privacy still depends on the workflow you choose.
Use this page as the canonical distinction between local dictation, cloud speech recognition, cloud AI providers, local AI providers, web search, and app operations.
Short Version
| Workflow | What can leave your device |
|---|---|
| Local dictation | No audio is sent to a cloud STT provider for transcription. |
| Cloud dictation | Audio or audio-derived data is sent to the selected STT provider. |
| Cloud AI processing | Clipboard text, selected text, images, instructions, and prompt context needed for the request can be sent to the selected AI provider. |
| Local AI processing | Requests can stay on your machine or local network, depending on your Ollama or LM Studio setup. |
| Web-grounded answers | The query and needed context are sent to the search-capable provider you selected. |
| Updates, downloads, activation | MachinesFluent may contact app infrastructure or third-party services needed for those operations. |
No compliance claim
These docs explain data flow. They do not claim that MachinesFluent is HIPAA-certified, enterprise-certified, or approved for any regulated workflow. If you work with regulated data, validate your own requirements before using any tool.
Local Dictation Path
When a local speech engine is selected:
- Your microphone audio is captured by MachinesFluent.
- The selected local model transcribes it on your computer.
- The transcript is inserted into the focused app or shown in the app workflow.
- If recording history is enabled, the transcript may be stored locally according to your settings.
This path avoids sending audio to a cloud speech-to-text provider. It does not prevent later workflows from sending text to an AI provider if you trigger them.
Cloud Speech Path
When a cloud speech engine is selected:
- MachinesFluent captures audio for the dictation.
- The audio or required audio-derived data is sent to the selected speech provider.
- The provider returns a transcript.
- MachinesFluent inserts or uses the transcript.
The selected provider's terms govern provider-side handling.
AI Provider Path
When you run an AI workflow, MachinesFluent sends the content needed for that workflow to the selected provider or local endpoint.
That content can include:
- clipboard text
- selected text
- copied images or screenshots
- spoken instructions after transcription
- saved prompt content
- system info or context files you configured
Use local providers such as Ollama or LM Studio when you want more local control. Use cloud providers when you accept that the request is leaving your device for that provider.
Web Search Path
Web-grounded answers require a search-capable provider. The query and relevant context must be sent to that provider so it can search or retrieve current information.
Do not use web search for secrets, private customer data, regulated records, credentials, or material you are not allowed to send to external services.
Recording History
MachinesFluent can keep a local history of dictated text and voice-command transcripts. Use history settings that match your risk level:
- lower limits for sensitive work
- higher limits for personal notes or low-risk drafting
- delete old entries when they are no longer useful
Practical Privacy Checklist
Before dictating or processing sensitive material, confirm:
- Which speech engine is active.
- Whether the speech engine is local or cloud.
- Whether AI processing is enabled for the workflow.
- Which AI provider or local endpoint is active.
- Whether recording history is enabled.
- Whether your organization permits the chosen provider.