Call Summaries

Call Summaries

Every AI Agent call generates a full transcript and a short AI-written summary. The summary is the piece most teams actually read — a two-to-four-sentence recap of why the caller called, what was discussed, and how it ended. This article explains what the summary captures, where it shows up, and how to tailor it for your workflow.

What a Summary Contains

By default, the summary answers three questions in plain language:

  • Why did they call? The caller's purpose or request.
  • What was exchanged? Key details — names, dates, phone numbers, addresses, stated preferences.
  • How did it end? Outcome — booked, transferred, left a message, escalated, hung up.

The summary is not a full transcript. It is designed to be scannable on a phone in under ten seconds. If you need the word-by-word detail, the full transcript is available in the portal or webhook.

Where the Summary Shows Up

The same summary is delivered wherever you have post-call actions configured:

  • Email — the body of the call recap email.
  • Webhook — the summary field in the webhook payload.
  • Post-call Integrations — Follows the Advanced options on the Integrations page.
  • Call logs in the portal — shown inline alongside the call record.

You set up what you want once; the same summary text flows through each channel.

Examples

Example Email:

Call Summary - Bob

Call Details

Receptionist: Bob
Extension: 8501
Caller: 15555551234
Start Time: 2026-04-20 02:34:57 PM EDT
End Time: 2026-04-20 02:35:22 PM EDT
Duration: 0 minutes, 25 seconds

Conversation Summary

Customer called Calldacity and spoke with AI Agent Bob regarding a billing inquiry. Agent confirmed the customer's request to be connected with the Billing department. Customer was transferred to the Billing department.

Example Webhook Payload

If you are wiring the agent into Zapier, Make, n8n, or a custom endpoint, here is the shape of the JSON the agent POSTs after every call. The call_data.call_summary field is the same summary described above; transcription is the full turn-by-turn log.

{
"event_type": "call_end",
"timestamp": "2026-04-21T12:36:10.500445Z",
"call_data": {
"call_id": "ai_usage_5329",
"usage_id": 5329,
"start_time": "2026-04-21T12:35:28.585122",
"end_time": "2026-04-21T12:36:06.560710",
"duration_seconds": 37,
"duration_minutes": 0.62,
"caller_number": "7045551234",
"call_summary": "# Call Summary\n\nThe caller contacted Generic Phones to inquire about the company's services. The agent provided an overview of Generic Phones' offerings, explaining that they provide cloud-based communication solutions including business phone systems, video conferencing, and customer integrations for businesses of various sizes. The caller ended the call without requesting additional information or scheduling any follow-up.",
"transcription": "Agent: Thank you for calling Generic Phones. You have reached Sumner's virtual assistant, how may I direct your call?\nCustomer: Yeah. I'm just calling to learn about what you guys do.\nAgent: Generic Phones is a provider of cloud-based communication solutions specializing in unified communications as a service. They offer services such as business phone systems, video conferencing, and customer integrations, all designed to improve communication and collaboration for businesses of various sizes. Would you like information on any specific service?\nCustomer: Cool, thanks. Bye.\nAgent: Thank you for calling Generic Phones. Have a great day. Goodbye.",
"recording_url": "https://recordings.calldacity.com/aiagentrecordings/111111111/sumnersawesomephones/SumnerAid/2026/04/21/5329.wav",
"trial_usage": true
},
"receptionist_data": {
"receptionist_id": 411,
"receptionist_name": "Bob",
"extension": "6301",
"reception_type": "basic"
},
"metadata": {
"webhook_version": "1.0",
"delivery_attempt": 1,
"receptionist_integration_id": 376
}
}

Notes on the fields you will use most often:

  • call_data.call_summary — the AI-written recap. Returned as Markdown (it typically begins with Call Summary). Render it as Markdown in your destination, or strip the heading if your tool cannot.
  • call_data.transcription — the full turn-by-turn transcript as a single newline-separated string. Each line is prefixed with Agent: or Customer:.
  • call_data.caller_number — the caller's number as presented on the call. May be a full E.164 number for outside callers or a short internal extension for on-system calls.
  • call_data.recording_url — an URI pointing at the stored recording. May be null if recording is disabled.
  • call_data.trial_usage — true for calls made against a trial account, false for production usage. Useful for filtering test traffic out of CRM automations.
  • call_data.start_time / end_time — ISO 8601 timestamps in UTC with microseconds. Note these are returned without a trailing Z; the outer timestamp field does include the Z. Treat both as UTC.
  • receptionist_data.extension — the extension the agent answers on. Handy for routing logic when the same webhook serves multiple agents.
  • receptionist_data.reception_type — basicai_receptionist, or the specific agent type. Lets downstream automations branch on agent kind.
  • metadata.delivery_attempt — starts at 1. If the agent retries delivery after a 5xx response, this increments.

Privacy and Retention

Summaries and transcripts are stored with the call record. Retention follows your account's overall data retention settings. If you have compliance requirements that exclude storing voice transcripts long-term, contact support to discuss redaction options for specific fields.

Troubleshooting

  • Summary is too generic. The default prompt is generic by design. If you need something specific you can use the webhook to get the full transcript or use the Collect Information and Execute Actions states to create custom alerts. 
  • Summary missed a detail. If the information was actually spoken on the call, check the full transcript — the summary drops detail for length.

Still have questions? Submit a ticket and we will help you tune your summary format.
    • Related Articles

    • Call Recordings, Transcripts & Audit Logs

      Every AI Agent call produces three artefacts you can review later: a full audio recording, a turn-by-turn transcript, and an AI-generated summary. Together they let you audit what the agent said, review edge cases, train staff on common call types, ...
    • OneCloud Professional Contact Center - Basic Tab Call Queue Configuration

      ?️ Basic Tab – Call Queue Setup Use the Basic tab to configure core settings for your call queue. These settings help define how your call queue functions and how it's tracked in reports. ? Name & Extension Set the name of the queue. A direct ...
    • Custom Call Center Agent Status | OneCloud Support Portal

      Call Center Supervisors can add custom agent statuses. Pre-defined statuses are broad descriptions such as "lunch", "break", and "available". By using a custom status, the Supervisor has increased granularity on call center reports. Note that custom ...
    • Zapier Post-Call Automation

      Zapier is the fastest way to turn an AI Agent call into action somewhere else — a row in a spreadsheet, a lead in a CRM, a message in Slack, a task in a project tracker — without writing any code. This article walks through the end-to-end setup using ...
    • 3CX: Call Queues

      This article covers Call Queues, their settings and different options. You will also learn how Queues work and what rights Agents and Queue Managers have in more detail: Call Queue Functionality Polling Strategies Queue Options Extension Login / ...