Use Spokenly with the Gateway¶
Spokenly is a voice dictation app for macOS, Windows, Linux, and iPhone. Each Spokenly mode can run your dictation through an AI model for cleanup, formatting, or translation. Point a mode at the Gateway to do that step in your region.
Prerequisites¶
- Spokenly installed
- An LLM API key
Connect Spokenly to the Gateway¶
- Open Spokenly Settings → Modes and pick a mode, or create a new one
- Set the mode's AI provider to the OpenAI-compatible option and enter:
| Field | Value |
|---|---|
| Base URL | https://llm.[your-region].amazee.ai/v1 |
| API key | Your LLM API key |
| Model | A fast model from Available Models, e.g. claude-4-5-haiku |
Replace [your-region] with your region code (e.g. ch103, us103).
- Set the mode's fallback provider to the same Gateway setup, or to none
- Save the mode and dictate something to test it
The fallback provider can bypass the Gateway
Each mode has a fallback provider. If it's set to Spokenly's built-in provider or to OpenAI, your text goes to that service, outside the Gateway and outside your region, whenever the main provider fails. Set the fallback to the Gateway or to none.
What goes through the Gateway¶
- The AI step of each mode: your AI instructions, cleanup, formatting, and translation
- Not transcription. Speech-to-text uses the mode's transcription model (on-device Whisper/Parakeet or a cloud provider)
If you use Spokenly Pro settings sync and turn on API key sync, your key is stored on Spokenly's servers. Leave that off if that's a concern.
Verify¶
Dictate with the mode you set up and check the My usage card on the AI keys page to confirm requests appear.
Troubleshooting¶
- AI processing not working
- Make sure the base URL, API key, and model ID are saved in the mode you're using.
- Transcription isn't billed to the Gateway
- This is expected. Voice-to-text uses its own models. The LLM API key is only used for the mode's AI step.
- Model not available
- Check Available Models for models available in your region.