Bot dashboard
The dashboard lists all your voice bots in a table with:- Bot name — display name
- Tag — environment label (staging, testing, or production)
- Bot ID — unique identifier
Actions
- Create bot — click “New agent” and provide a name
- Edit bot — click a row or use the three-dot menu
- Copy bot ID — copy the identifier to clipboard
- Delete bot — remove a bot with confirmation
Bot editor
The editor has two panels: the main configuration on the left and advanced settings on the right.Main configuration
| Field | Description |
|---|---|
| Agent intercom | Phone number for call transfers to a human agent |
| Greeting message | The first message the bot speaks on a call |
| System prompt | Instructions for the LLM defining the bot’s role |
| System instruction | Additional behavioral instructions |
| Handle user idle | Messages sent when the caller is silent (add multiple) |
| Functions | Callable functions the bot can trigger during a conversation |
| Departments | Named departments with phone numbers for transfers |
Language
Supported languages: English (IN), Tamil, Hindi, Telugu, Kannada, Malayalam, English (US).Text-to-speech (TTS)
| Setting | Options |
|---|---|
| Provider | Gemini or Google |
| Model | gemini-2.5-flash-tts, gemini-2.5-pro-tts |
| Voice | 30 voice styles (Aoede, Achernar, Algenib, etc.) |
| TTS prompt | Tone and style instructions (Gemini only) |
Speech-to-text (STT)
| Setting | Options |
|---|---|
| Model | latest_long, latest_short, chirp_2, chirp, telephony |
| Mute strategy | first_speech, mute_until_first_bot_complete, function_call, always |
LLM model
Choose from:gemini-2.5-flash, gemini-2.5-flash-lite, gemini-2.5-pro, gemini-2.0-flash-lite-001.
Voice activity detection (VAD)
Fine-tune how the bot detects speech:| Parameter | Range | Default |
|---|---|---|
| Confidence | 0.0–1.0 | 0.8 |
| Start secs | 0.0–2.0 | 0.3 |
| Stop secs | 0.0–3.0 | 0.8 |
| Min volume | 0.0–1.0 | 0.6 |
Additional settings
- Music — enable hold music
- Fetch data — fetch caller data before the call
- Noise filter — filter background noise
Webhook
Configure a webhook to receive call events:- URL — your endpoint
- Method — POST or GET
- Headers — custom headers (key-value pairs)
- Params — query parameters (key-value pairs)
Phonebook
Manage a contact directory the bot can reference:- Upload CSV — replace or append contacts
- Lookup — search by phone number
- Add entry — add a single contact
- Delete — clear the entire phonebook
Testing
Click Test in the editor to open a voice tester. This connects via WebSocket to simulate a live call with the bot.Import and export
- Download — export the bot configuration as a JSON file
- Import — upload a JSON file to apply a configuration