An outbound AI calling campaign is a programmatically triggered sequence of phone calls — made by an AI voice agent or automated system — at scale, without manual dialing. The infrastructure layer handles number provisioning, call initiation, answering machine detection, dynamic call flow execution, and post-call data delivery to your CRM or data warehouse.

Most teams building this on legacy CPaaS platforms spend hours across dashboards, credential panels, and SDK setup before a single call fires. On Vobiz, the full stack — DID provisioning, AMD, XML call flow, and hangup webhooks — is configured through a single REST API with two auth headers on every request. This guide walks through all four steps with real code.

Vobiz uses an event-driven webhook architecture for call flow control. The sequence works like this:

1. Fire — Your backend fires a POST to the Vobiz Make Call API with a from, to, and answer_url.
2. Dial — Vobiz dials the destination number.
3. Connect — When the call connects, Vobiz sends a POST to your answer_url. Your server responds with Voice XML instructions — play audio, stream to an AI agent, collect DTMF input, or hang up.
4. AMD — If you've enabled AMD (Answering Machine Detection), Vobiz runs acoustic analysis in parallel and fires a callback to your machine_detection_url with a Machine: true/false flag.
5. Hangup — When the call ends, Vobiz fires your hangup_url with full timestamps — StartTime, AnswerTime, EndTime — ready to write to your CRM.

Each webhook cycle is stateless. Your application tracks call state using the CallUUID Vobiz sends with every event.

• A Vobiz account — sign up and get ₹25 in free credit to start
• Your Auth ID and Auth Token from the Vobiz dashboard — required on every API request as X-Auth-ID and X-Auth-Token headers
• A publicly reachable HTTPS endpoint for your answer_url, machine_detection_url, and hangup_url (use ngrok locally during development)
• A DID number in your target country (provisioned in Step 1 below)

Authentication model: Every request to https://api.vobiz.ai/api/v1 requires two headers:

X-Auth-ID: YOUR_AUTH_ID
X-Auth-Token: YOUR_AUTH_TOKEN
Content-Type: application/json

Before you can place a call, you need a caller ID — a DID number that Vobiz assigns to your account. Vobiz supports DID provisioning across 130+ countries, including India's 140, 1600, and 92 series numbers.

Browse available numbers first:

curl -X GET https://api.vobiz.ai/api/v1/Account/{auth_id}/inventory/numbers \
  -H "X-Auth-ID: YOUR_AUTH_ID" \
  -H "X-Auth-Token: YOUR_AUTH_TOKEN"

Purchase the number:

curl -X POST https://api.vobiz.ai/api/v1/Account/{auth_id}/numbers/purchase-from-inventory \
  -H "X-Auth-ID: YOUR_AUTH_ID" \
  -H "X-Auth-Token: YOUR_AUTH_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "e164": "+919876543210"
  }'

Response (200 OK):

{
  "message": "Number purchased successfully",
  "number": {
    "e164": "+919876543210",
    "country": "IN",
    "region": "Karnataka",
    "status": "active",
    "voice_enabled": true,
    "monthly_fee": 1,
    "currency": "INR"
  }
}

The number is active immediately. voice_enabled: true confirms it's ready for outbound calls — no waiting period, no support ticket.

Note for India outbound campaigns: If you're dialing Indian mobile numbers at scale, you will need a 140-series or 1600-series number and DLT registration for TRAI compliance.

This is the core API call. Everything — the caller ID, destination, call flow webhook, AMD config, and hangup callback — is set in a single POST request.

Endpoint: POST https://api.vobiz.ai/api/v1/Account/{auth_id}/Call/

Path casing matters: The URL requires a capital A in Account, a capital C in Call, and a trailing slash. Lowercase or missing slash returns a 401.

curl -X POST https://api.vobiz.ai/api/v1/Account/{auth_id}/Call/ \
  -H "X-Auth-ID: YOUR_AUTH_ID" \
  -H "X-Auth-Token: YOUR_AUTH_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "from": "+919876543210",
    "to": "+919123456789",
    "answer_url": "https://yourapp.com/answer",
    "answer_method": "POST",
    "hangup_url": "https://yourapp.com/hangup",
    "hangup_method": "POST",
    "machine_detection": "hangup",
    "machine_detection_url": "https://yourapp.com/amd-callback",
    "machine_detection_method": "POST",
    "machine_detection_time": 5000,
    "time_limit": 3600
  }'

Store the request_uuid from the response — this is your CallUUID and will appear in every subsequent webhook for this call.

AMD: Sync vs. Async Modes

Machine Detection is not a separate endpoint. It's configured via parameters on the same call request.

Recommended for campaigns: Use machine_detection: "hangup" with machine_detection_url together. Vobiz hangs up on machines and still fires your callback so you can log the outcome.

When a human answers the call, Vobiz hits your answer_url and expects Voice XML back. Your server generates XML dynamically — personalize it per recipient using the From, To, and CallUUID params Vobiz sends with the webhook.

Option A: Spoken IVR with DTMF collection

Use <Speak> and <Gather> to deliver a message and collect a keypress:

<?xml version="1.0" encoding="UTF-8"?>
<Response>
  <Gather numDigits="1" timeout="10" action="https://yourapp.com/gather-result" method="POST">
    <Speak>Hi, this is a call from Acme. Press 1 to speak with our team, or press 2 to be removed from our list.</Speak>
  </Gather>
  <Speak>We didn't receive your input. We'll try again soon. Goodbye.</Speak>
  <Hangup/>
</Response>

Option B: Stream directly to an AI voice agent

Use <Stream> to fork live audio to a WebSocket server — your AI agent (Vapi, Retell AI, ElevenLabs, LiveKit, Pipecat) handles the conversation in real time:

<?xml version="1.0" encoding="UTF-8"?>
<Response>
  <Stream streamTimeout="86400" keepCallAlive="true">
    wss://your-ai-agent.com/audio-stream
  </Stream>
</Response>

Use the Vobiz XML Builder to compose and preview XML without writing it by hand.

When the call ends, for any reason, Vobiz fires a POST to your hangup_url. No XML response is expected. This is your data capture event: write call outcome, duration, and timestamps directly to your CRM.

{
  "From": "+919876543210",
  "To": "+919123456789",
  "CallUUID": "97ceeb52-58b6-11e1-86da-77300b68f8bb",
  "Direction": "outbound",
  "Event": "Hangup",
  "CallStatus": "completed",
  "StartTime": "2026-06-23 10:00:00",
  "AnswerTime": "2026-06-23 10:00:06",
  "EndTime": "2026-06-23 10:02:34",
  "STIRAttestation": "A",
  "stir_verification": "Verified"
}

Minimal Python handler:

from flask import Flask, request
import your_crm_client as crm

app = Flask(__name__)

@app.route('/hangup', methods=['POST'])
def hangup():
    data = request.form
    crm.update_call_record(
        call_uuid=data.get('CallUUID'),
        status=data.get('CallStatus'),
        start_time=data.get('StartTime'),
        answer_time=data.get('AnswerTime'),
        end_time=data.get('EndTime'),
        direction=data.get('Direction')
    )
    return '', 200

Return a 200 immediately — Vobiz does not wait for a response body, but a slow or failing response may cause retry behavior.

For campaign-scale outbound, Vobiz supports up to 1,000 destinations in a single API call. Separate numbers with <:

curl -X POST https://api.vobiz.ai/api/v1/Account/{auth_id}/Call/ \
  -H "X-Auth-ID: YOUR_AUTH_ID" \
  -H "X-Auth-Token: YOUR_AUTH_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "from": "+919876543210",
    "to": "+919111111111<+919222222222<+919333333333",
    "answer_url": "https://yourapp.com/answer",
    "hangup_url": "https://yourapp.com/hangup",
    "machine_detection": "hangup",
    "machine_detection_url": "https://yourapp.com/amd-callback"
  }'

Each destination gets its own call leg, its own CallUUID, and independent webhook delivery. Make sure your server handles parallel webhook processing.

Vobiz also ships official SDKs for Python, Node.js, Ruby, Go, and C#.

If your campaign targets Indian mobile numbers, TRAI's DND and DLT regulations apply before you can dial at scale:

• 140-series numbers are required for transactional and service calls (OTP, reminders, alerts). See how to acquire 140/160-series numbers on Vobiz.
• DLT registration — your calling header and call flow template must be pre-approved on the DLT platform. Vobiz's DLT registration guide covers the submission process.
• Calling hours — TRAI restricts outbound calls to 9am–9pm in the recipient's local time. Full rules are in the India Compliance docs.
• KYC — Indian accounts require GST, PAN, and Aadhaar verification before production calling volumes are unlocked.

Vobiz bills in INR and issues GST-compliant invoices for Indian accounts.

The developer experience difference is primarily in three places: AMD is a first-class parameter on Vobiz (not an add-on), bulk dialing is native, and India compliance is documented explicitly rather than left to the developer to interpret TRAI rules.