Creating an outbound campaign
Set up your Voice AI agent
Go to AI Agents → Voice AI and create or select an agent. Configure its personality, knowledge base, and call script. See Creating agents.
Select your contact list
Choose the contacts or smart list to call. You can filter by tags, custom fields, or other criteria.
Configure call settings
- Caller ID — select the phone number to display
- Call script — assign the Voice AI agent to handle the conversation
- Max attempts — how many times to retry if the contact does not answer
- Retry interval — time between retry attempts
Schedule the campaign
Set the start date, end date, and calling hours. Only calls within the allowed hours will be placed.
Compliance requirements
| Requirement | Details |
|---|---|
| Prior express consent | You must have documented consent to call each contact |
| Time-of-day restrictions | Calls must be placed between 8:00 AM and 9:00 PM in the contact’s local timezone |
| Caller ID | Must display a valid, callable phone number |
| Do-Not-Call list | Scrub your list against the National Do-Not-Call Registry |
| Opt-out mechanism | Contacts must be able to say “stop” or press a key to opt out during the call |
| AI disclosure | Some jurisdictions require disclosing that the caller is an AI agent |
Call scripts and AI behavior
Your Voice AI agent follows the instructions and knowledge base you configure:- System prompt — defines the agent’s personality, goals, and conversation flow
- Knowledge base — provides information the agent can reference during calls
- Actions — what the agent can do (book appointments, transfer calls, collect information)
- Guardrails — topics or questions the agent should not address
Monitoring the outbound dashboard
The campaign dashboard shows real-time metrics:| Metric | Description |
|---|---|
| Calls placed | Total calls attempted |
| Connected | Calls where the contact answered |
| Voicemail | Calls that went to voicemail |
| No answer | Calls with no pickup and no voicemail |
| Completed | Calls where the AI finished the conversation |
| Transferred | Calls escalated to a human agent |
| Appointments booked | Appointments created during calls |
| Average duration | Mean call length |
Call logs and transcripts
Every outbound call generates:- Call recording — audio of the full conversation
- Transcript — AI-generated text transcript
- Summary — brief overview of what happened on the call
- Outcome — result (appointment booked, callback requested, not interested, etc.)
- Contact activity — the call appears on the contact’s timeline
Cost per call
| Component | Cost |
|---|---|
| Outbound call minutes | Standard per-minute outbound rate (~$0.014/min US) |
| Voice AI processing | AI credits per minute of AI conversation |
| Total estimated cost | ~$0.05–0.15 per minute depending on AI model |
Costs vary based on call duration, AI model selected, and destination country. Monitor usage in Settings → Billing → Usage.
Best practices
Start with a small test batch
Start with a small test batch
Call 20–50 contacts first to test the script, measure outcomes, and refine before scaling.
Optimize call times
Optimize call times
Schedule calls during business hours when contacts are most likely to answer. Avoid early morning and late evening.
Keep the script conversational
Keep the script conversational
AI agents sound more natural with short, conversational prompts rather than long scripts. Let the AI adapt to the conversation.
Set up human escalation
Set up human escalation
Always configure a path to transfer to a live agent if the contact requests one or the conversation goes beyond the AI’s scope.