Provider Management
Create and configure the global pool of AI providers (Chat, Embedding, STT, TTS) available across all bots. Individual bots then activate providers from this pool on their Enabled Providers page.
The Provider Management page is the tenant-level administration for all AI providers. Providers created here are available as a shared pool that individual bots can then activate and assign roles to (see Enabled Providers).
Providers are grouped into four sections: Chat, Embedding, STT (Speech-to-Text), and TTS (Text-to-Speech). Each row shows the provider's icon, title, last-modified date, deployment model name, a System badge if applicable, and a status indicator.
Status Indicators
| Provider Type | Indicator | Meaning |
|---|---|---|
| Chat / Embedding | Green checkmark | Provider is reachable and healthy |
| Chat / Embedding | Yellow triangle | Provider is unreachable or misconfigured |
| STT / TTS | Green checkmark | Synced with the phone server |
| STT / TTS | Hourglass | Sync pending |
| STT / TTS | Red × | Sync failed |
| STT / TTS | Yellow triangle | Configuration changed, re-sync needed |
Creating a Provider
Click Create to open the creation dialog. Fill in:
- API type — Chat, Embedding, STT, or TTS
- Provider type — the underlying service:
- Chat / Embedding: OpenAI, Azure OpenAI, Together AI, fireworks.ai, Custom
- STT: Google Speech-to-Text, Azure Speech Services, botario Speech
- TTS: Google Text-to-Speech, Azure Speech Services, ElevenLabs, botario
- General settings — Title, API Key, Endpoint URL, Model name (or Voice name for TTS)
Depending on the provider type, additional options appear:
- Azure (Chat / Embedding): API version; optionally Microsoft Entra ID authentication (tenant ID, client ID, client secret, scope)
- Chat: Reasoning Model toggle with a free-text effort level (e.g. none, minimal, low, medium, high, xhigh — supported values depend on the model); Custom parameters (JSON, passed directly to LiteLLM)
- Embedding with type
custom: Botario Embedding Model flag - STT: Language code, region (Azure), profanity filter, punctuation toggles, wait-time-after-speech
- TTS: Language code, speaking rate, pitch, gender (Google), voice model (ElevenLabs, botario), similarity boost and stability (ElevenLabs)
After filling in the form, use Test Settings to validate the credentials before saving. For chat providers, the test also shows which capabilities the model supports (normal chat, response model format, tool calling).
Editing a Provider
Click any row to open the edit dialog. For system providers (marked with a System badge), the provider type, API type, URL, and deployment name are read-only — only the title, API key, reasoning model flag, and custom parameters can be changed. For custom providers, all fields except type and API are editable.
Deleting a Provider
Open a provider and click Delete. If the provider is currently assigned to any bot (as Default, Orchestrator, Autocomplete, or any enabled role), deletion is blocked and the affected bots are listed. Remove the provider from those bots on their Enabled Providers page first.