After-Call Webhook

When a call session ends, Agni dispatches a call.completed webhook event to any URLs configured in your agent’s Webhook Settings. This delivers the full call record including transcripts, AI analysis, and metadata.

Configure webhook URLs in the agent builder’s Webhook Settings panel, or via the Agent API webhook_urls field.

Event Details

Field	Value
Event name	`call.completed`
Method	`POST`
Content-Type	`application/json`
Trigger	Automatically dispatched when a call session ends

Payload Structure

string

Event type identifier. Always "call.completed" for this webhook.

string

UUID of the organization that owns this call.

object

Full call session data.

Show data

string

Campaign ID if the call was part of an outbound campaign. null for inbound/web calls.

string

Contact ID if the caller is a known contact in your CRM.

string

Phone number of the external party in E.164 format.

string

Call outcome status. Values: completed, failed, no_answer, busy, voicemail.

integer

Total call duration in seconds.

string

Unique session ID for this call.

integer

Attempt number (for campaign retries). 1 for first attempt.

string

AI-generated summary of the conversation. Powered by the agent’s analysis_summary_prompt.

string

URL to the full call recording audio file (MP3).

string

The phone number that initiated the call (your agent’s number for outbound, caller’s number for inbound).

string

The phone number that received the call.

string

Name of the caller (if known from contacts).

string

Email of the caller (if known from contacts).

string

Name of the AI agent that handled the call.

string

Call channel. Values: voice, web, sip.

string

Why the call ended. Values: customer_hangup, agent_hangup, error.

number

Total cost of the call in USD.

integer

Average response latency in milliseconds.

string

ISO 8601 timestamp when the call started.

string

ISO 8601 timestamp when the call ended.

string

ISO 8601 timestamp when the session record was created.

object

AI-generated analysis of the call.

Show post_call_analysis

string

Caller sentiment. Values: positive, neutral, negative.

string

Call outcome classification (e.g. meeting_booked, not_interested, callback_requested).

boolean

Whether the agent achieved its defined objective.

string

AI-suggested next action (e.g. “Send follow-up details via email”).

object[]

Full conversation transcript, ordered chronologically.

Show transcripts[]

string

Unique transcript entry ID.

integer

Milliseconds from the start of the call when this message was spoken.

string

Speaker role. Values: agent, user.

object

Message content.

Show message

string

The spoken text.

string

Content format. Typically "text".

string

ISO 8601 timestamp.

Example Payload

{
  "event": "call.completed",
  "org_id": "1268c1f0-19f3-47db-aefb-c16a7c3ace6e",
  "data": {
    "campaign_id": "019d2a1b-4c5e-7f8a-9b0c-1d2e3f4a5b6c",
    "contact_id": "019d2a1b-5d6f-7a8b-9c0d-2e3f4a5b6c7d",
    "phone": "+14155550100",
    "status": "completed",
    "duration_sec": 125,
    "call_session_id": "019d2b3c-8e9f-7a0b-1c2d-4e5f6a7b8c9d",
    "attempt": 1,
    "summary": "Customer called about their order #4521. Agent confirmed the order shipped yesterday and provided the tracking number. Customer was satisfied.",
    "recording_url": "https://storage.agniai.com/rec/019d2b3c-8e9f.mp3",
    "caller_number": "+18881234567",
    "callee_number": "+14155550100",
    "caller_name": "John Doe",
    "caller_email": "john@example.com",
    "agent_name": "Support Agent",
    "channel": "voice",
    "disconnect_reason": "customer_hangup",
    "cost_total": 0.42,
    "call_latency_ms": 150,
    "started_at": "2026-03-24T18:30:00Z",
    "ended_at": "2026-03-24T18:32:05Z",
    "created_at": "2026-03-24T18:30:00Z",
    "post_call_analysis": {
      "sentiment": "positive",
      "disposition": "issue_resolved",
      "goals_met": true,
      "next_steps": "No follow-up needed. Customer issue fully resolved."
    },
    "transcripts": [
      {
        "id": "t-001",
        "timestamp_ms": 0,
        "role": "agent",
        "message": { "content": "Hello! Thanks for calling Acme Support. How can I help you today?", "format": "text" },
        "created_at": "2026-03-24T18:30:00Z"
      },
      {
        "id": "t-002",
        "timestamp_ms": 3200,
        "role": "user",
        "message": { "content": "Hi, I wanted to check on my order number 4521.", "format": "text" },
        "created_at": "2026-03-24T18:30:03Z"
      },
      {
        "id": "t-003",
        "timestamp_ms": 5800,
        "role": "agent",
        "message": { "content": "Of course! Let me look that up for you. Order 4521 shipped yesterday and the tracking number is TRK-98765. It should arrive by Friday.", "format": "text" },
        "created_at": "2026-03-24T18:30:05Z"
      }
    ]
  }
}

Handling Webhooks

Respond with 200

Your endpoint must return a 200 HTTP status code to acknowledge receipt. If Agni doesn’t receive a 200, it will retry the webhook.

Verify the org_id

Always check that the org_id in the payload matches your organization to prevent processing webhooks from other sources.

Process asynchronously

Return 200 immediately and process the data in a background job. Don’t make the webhook wait for your database writes or API calls.

Store the call_session_id

Use call_session_id as a unique key to deduplicate webhook deliveries in case of retries.

Introduction

Agent API

Organization API

Agent Tool API

Phone Number & Telephony API

Campaign API

Contact API

Calling API

Appointment API

GHL Integration API

Cal.com Integration API

RAG Knowledge Base API

Webhooks

Health API

After-Call Webhook

Event Details

Payload Structure

Example Payload

Handling Webhooks

​Event Details

​Payload Structure

​Example Payload

​Handling Webhooks

Event Details

Payload Structure

Example Payload

Handling Webhooks