> For the complete documentation index, see [llms.txt](https://docs.maiagent.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.maiagent.ai/maiagent-user-guide/en/serve/voicecs/call-qa.md).

# Voice Call Quality Assurance

Voice call quality inspection example: after a customer service call is transcribed to text, an Agent checks scripts, sentiment, and compliance risks against the SOP and produces a quality inspection

This chapter explains how to build an intelligent voice call quality inspection Agent. Combined with Automatic Speech Recognition (ASR) technology, it transcribes calls between customer service representatives and customers into text, and the Agent then automatically reviews service quality. This helps supervisors quickly understand how customer service is performing and avoid missing potential service flaws or compliance risks. The Agent needs the following core capabilities:

1. **Service standard compliance check**: Based on the company's established customer service SOP, determine item by item whether the representative completed the required scripts, such as self-introduction, customer identity verification, and closing courtesy.
2. **Sentiment and attitude analysis**: Analyze the representative's tone and word choice to identify situations such as impatience, arguing, or unprofessional behavior, and track changes in the customer's sentiment.
3. **Compliance and prohibited phrase detection**: Automatically check whether the call contains promissory language, sensitive words, or scripts that violate regulations, reducing legal compliance and complaint risks.
4. **Structured inspection output**: Output inspection results in structured fields, making it easy to import them into back-end systems or BI reports and helping supervisors track and analyze service quality.

Fully automated inspection by the Agent greatly reduces the burden of manual spot checks. Instead of sampling only a small number of calls as before, every call can now be inspected in real time, helping enterprises spot service gaps and make improvements right away.

### System Prompt <a href="#system-prompt" id="system-prompt"></a>

```markdown
# Role
You are a call quality inspection assistant for an enterprise customer service center. The input is the transcript of a call between a representative and a customer

# Inspection Items
1. Opening: Did the representative give a complete self-introduction (company name, name/employee ID)
2. Identity verification: Did the representative verify the caller's identity as required
3. Service attitude: Was the representative's tone friendly, with no impatience or arguing
4. Prohibited phrase detection: Did any promissory, aggressive, or non-compliant language appear
5. Closing courtesy: Did the representative confirm the customer's issue was resolved and say goodbye proactively

# Output Restrictions
1. Each inspection result may only be "Pass", "Fail", or "N/A"
2. If an item fails, add the original excerpt and the reason in the "Details" field
3. Put "None" for any field that cannot be determined

# Output Format
Output the text directly, without <template> and </template>

<template>
Opening: Pass/Fail/N/A
Identity verification: Pass/Fail/N/A
Service attitude: Pass/Fail/N/A
Prohibited phrase detection: Pass/Fail/N/A
Closing courtesy: Pass/Fail/N/A
Details: XXX
Overall recommendation: XXX
</template>
```

### Voice Conversation Transcript <a href="#voice-conversation-transcript" id="voice-conversation-transcript"></a>

```
Representative:
Hello, this is the MaiAgent Customer Service Center. I'm Mei, employee ID A123. How may I help you?

Customer:
Hello, the amount on my bill last month is wrong. Why was I charged an extra five hundred dollars?

Representative:
I'm very sorry for the trouble. To protect your rights, may I first verify your identity? Please provide your member number or registered mobile phone number.

Customer:
My phone number is 0912-345-678.

Representative:
Thank you, I've found your account. I can see that last month's bill does include a five-hundred-dollar value-added service fee, generated by the "Family Protection Plan" you activated on our website on November 5.

Customer:
I never activated that plan! Are you just charging people randomly?

Representative:
I understand your concern. I will immediately pull the activation record and report back. A specialist will contact you within 24 hours with a resolution. May we reach you at this mobile number?

Customer:
Yes, thank you for your help.

Representative:
Alright, I've created a case for you. Is there anything else I can help you with?

Customer:
No, thank you.

Representative:
Thank you for calling the MaiAgent Customer Service Center. Have a great day, goodbye.
```

### Inspection Result <a href="#qa-result" id="qa-result"></a>

```markdown
Opening: Pass
Identity verification: Pass
Service attitude: Pass
Prohibited phrase detection: Pass
Closing courtesy: Pass
Details: The representative gave the full company name and employee ID, and asked the customer for a mobile phone number to verify identity as required. When the customer raised doubts, the representative stayed calm, did not argue or use promissory language, and proactively created a follow-up case.
Overall recommendation: This call meets service standards. When responding to customer concerns, it is recommended to explain the specific process for pulling records in more detail to strengthen customer trust.
```


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.maiagent.ai/maiagent-user-guide/en/serve/voicecs/call-qa.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
