Voice AI
Create and configure a Voice Agent
Build a focused multilingual phone agent with a controlled opening, clear scope and safe escalation path.
Before you start
- Define inbound or outbound use, target language and call objective.
- Prepare approved knowledge and escalation rules.
- Confirm Voice AI allocation in the account.
Step by step
1
Create the agent
Open Voice AI, create an agent and use an internal name that identifies the business and call purpose.
2
Choose model, language and voice
Select supported options that fit answer quality, latency, accent and usage requirements.
3
Write the first sentence
Make the opening identify the business and purpose without sounding deceptive or overly long.
4
Add the voice prompt
Define scope, qualification, booking, prohibited claims, transfer conditions and how the agent handles silence or uncertainty.
5
Choose deployment
Decide whether the agent will use a phone number, Voice Widget or another supported call flow, then test before launch.
Verify the result
- The opening is clear in the selected language.
- The agent does not invent facts or call outcomes.
- Transfer and ending behaviour work predictably.
Common issues
- Latency feels high: Shorten unnecessary prompt and knowledge context and test the supported model options.
- Voice pronunciation is poor: Test another supported voice or adjust approved wording and normalization.
- Agent speaks over callers: Review interruption, silence and speech-detection settings.
Important: Voice AI is available from FADA GROWTH. Phone numbers, call providers and regional availability can add separate requirements and costs.
Next: Configure call settings
Continue with the next recommended FADA guide.