Privacy

No account, no telemetry — only what dictation actually needs.

How it works

Where your voice and text go

1. Audio goes to Groq

   Your recording is sent straight from your device to Groq, using your own API key.

2. Text comes back to your device

   Groq returns the transcript, the app cleans it up and inserts it right where your cursor is.

3. Our server isn't in the loop

   There's no Kama Type server in this chain — it's just your device and Groq.

Selected text

The one feature that reads more than your voice

The Windows version has a separate setting called “Selected Text Context”. It is off by default: until you turn it on, the app looks neither at the foreground window nor at your clipboard — only your recording goes to Groq.

Turn it on and, the moment you finish dictating, the app simulates Ctrl+C in the foreground window, picks up the selected fragment and sends it to Groq along with the transcript — under your own key, by the same route, with no server of ours involved. The app restores whatever was in your clipboard before. If nothing is selected, the clipboard doesn't change and there is nothing to send — no fragment leaves your device.

You switch it off where you switched it on: Settings → LLM → “Selected Text Context”. Keep it off when someone else's or sensitive documents are on screen: Ctrl+C picks up exactly what is selected at that moment.

Storage

Where everything lives

• Dictation history and statistics live in a local database on the device itself. There is no copy on a server, because there is no server.
• On Windows the transcription key is kept in the system credential store, not in the settings file.
• The dictionary and snippets are on the device too, and start empty.
• The app updates only by hand: there is no auto-update, and it never downloads new code by itself.

What's missing

What we don't collect or offer

• An account
• Registration
• Telemetry
• In-app analytics
• Cloud dictation history
• A subscription

Network whitelist

Where the app actually connects

By default, exactly two addresses — the app contacts nothing else. The second one is only called when you submit the feedback form yourself. In the desktop version you can switch recognition from Groq to OpenAI, in which case audio goes to api.openai.com; and for text clean-up you can enter an address of your own, including a local Ollama instance.

• api.groq.com
• feedback.kamatype.app

This site

Visit statistics

Site visits are counted from server logs, IP addresses are anonymized on aggregation. No cookies, no third-party tracker.

The legal wording lives in the privacy policy