Bot Settings

Call Pipeline
Voice (STT/TTS)
LLM Model
Knowledge Base
Pipeline Mode
Choose how audio is processed during calls

OpenAI Realtime

All-in-one voice-to-voice. STT, LLM & TTS handled internally by OpenAI. Lowest latency, most natural.

~200ms latency

Modular Pipeline

Mix & match STT, LLM, and TTS providers independently. Full control over each component.

Fully customizable
Call Settings
In Realtime mode, STT, LLM, and TTS are all handled by OpenAI's Realtime API. Switch to Modular to pick individual providers on the Voice and LLM tabs.
Speech-to-Text (STT)
Which provider converts caller speech to text
Text-to-Speech (TTS)
Which provider generates AI voice responses
Voice Selection
Choose the voice for AI responses (based on selected TTS provider)
Select a TTS provider above to see available voices
LLM Model
The AI model that powers conversation intelligence
Temperature
Controls creativity vs consistency. Lower = more focused, Higher = more creative.
Precise Creative
0.8
System Prompt Override
Custom instructions for the AI. Leave blank to use the auto-generated prompt from campaign data.
Knowledge Base
FAQ, objection handling, product info — the AI uses this to answer questions

Loading...