Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.talkturo.ai/llms.txt

Use this file to discover all available pages before exploring further.

Every call that flows through Talkturo produces a full text transcript of the conversation. Talkturo also runs an AI analysis pass after each call ends — generating a summary, a success evaluation, and any structured data your assistant was configured to extract. You can read all of this directly in the Call Logs detail view, or retrieve it via the API.

​
Opening a transcript

1

Go to Call Logs

Navigate to Call Logs in the left sidebar.
2

Find the call

Use the filter bar to locate the call you want to review — filter by assistant, campaign, phone number, or search by call ID.
3

Open the detail panel

Click the magnifying glass icon at the left of the row, or click anywhere on the row itself. The Call Detail panel slides open.
4

Select the Transcripts tab

In the tab bar, click Transcripts. The full conversation text appears in the panel.

​
Transcript format

The Transcripts tab displays the raw conversation text captured by the speech-to-text engine. The text preserves the natural flow of the conversation, including pauses and interruptions as they occurred. For a message-by-message view with speaker labels and timestamps, switch to the Messages tab. Each message entry shows:
  • Speaker role — assistant (the AI), user (the caller), or system (internal events).
  • Timestamp — The Unix time the message was recorded, displayed as a human-readable date and time.
  • Message text — What was said, shown in a monospace font.

​
AI-generated analysis

After a call ends, Talkturo automatically runs an analysis pass on the transcript and writes the results to the Analysis tab. This tab has four sections:
A plain-English paragraph written by the AI that describes what happened in the call — who the caller was, what they asked or discussed, and how the conversation concluded. The summary is generated from the full transcript, not just the last few messages.
A true or false value that indicates whether the call met the success criteria defined in your assistant’s configuration. Use this field to calculate your assistant’s completion rate across a campaign.
Structured JSON containing any information your assistant was configured to extract during the conversation — for example, a contact’s email address, appointment preference, or qualification status. This data feeds directly into CRM field updates.
Additional structured data captured by the assistant — call metadata, custom tags, or any extra context your assistant appended during processing.

​
Sentiment analysis

Talkturo’s AI analysis includes a sentiment signal embedded in the summary and success evaluation. While there is no separate sentiment score field, the analysis summary will describe the caller’s tone (e.g., “the caller expressed frustration” or “the caller responded positively”). Use the success evaluation field as a proxy for call sentiment at scale — a false evaluation on a completed call often correlates with a negative interaction.

​
How transcripts feed into CRM updates

After each call, the extracted Data from the Analysis tab is written back to the matching contact record in your CRM. If the assistant extracted an email address, that value populates the contact’s email field. If a custom field was targeted (e.g., lead_score), it is created or updated automatically. This pipeline runs without any manual action on your part. You can review exactly what was extracted by opening the Analysis tab for any call.

​
Retrieve a transcript via API

To fetch a call’s full details — including transcript, analysis, recording URL, and message log — use the following endpoint: 
GET /api/calls/{call_id}
Authorization: Bearer <token>
Example response (abbreviated) 
{
  "id": "row-uuid",
  "call_id": "call-uuid",
  "transcript": "Hello, this is Alex from Talkturo...",
  "call_analysis": [
    {
      "summary": "The caller inquired about pricing for the Pro plan...",
      "success_evaluation": true,
      "data": {
        "email": "alex@example.com",
        "interested_plan": "pro"
      },
      "metadata": null
    }
  ],
  "call_messages": [
    {
      "role": "assistant",
      "time": 1714000000000,
      "message": "Hello! How can I help you today?"
    },
    {
      "role": "user",
      "time": 1714000005000,
      "message": "Hi, I'd like to know about your pricing."
    }
  ]
}
The call_id in the URL is the full UUID shown at the top of the Call Detail panel, not the 8-character truncated version displayed in the table.