1. What is AI Quality Management (QM)?
AI Quality Management is a full-scale AI-powered quality inspection feature based on multimodal large models. It automatically analyzes call center recordings to achieve 100% coverage, addressing the blind spots of traditional sampling-based quality checks, and improving service quality and compliance.
For enabling this feature, please contact NXAI staff.
2. Product Features
2.1 AI QM Task List Page
This page displays all configured AI QM tasks.
- Supports filtering by task name and agent group

| Task Status | Description |
|---|---|
| Ready |
“Pause” button is disabled |
| In Progress |
“Start” button is disabled |
| Paused |
“Pause” button is disabled |
| Completed |
Both “Start” and “Pause” buttons are disabled |
2.2 Create AI QM Task

2.2.1 QM Configuration
| Field Name | Description |
|---|---|
| Task Name | Enter the task name |
| Task Description | Enter a description for the task |
| Agent Groups | Supports multi-selection of voice agent groups |
| Time Range | Cannot exceed 1 month (YYYY-MM-DD HH:mm:ss : YYYY-MM-DD HH:mm:ss) |
| Duration | Configurable between 30–600 seconds |
| Sampling Rate (%) | Percentage of recordings randomly selected from eligible calls |
If no voice call records exist for the selected agent groups, the system will prevent proceeding to the next step.
2.2.2 QM Strategy

| Field Name | Description |
|---|---|
| Prompt Content | Default pre-built prompt (editable), max 3000 characters |
| AI Score | Automatically scored based on predefined rules (not editable) |
| AI Comment | Max 200 characters, limited to 100 tokens for generation |
2.2.3 QM Budget

After configuration, the system calculates the required budget based on:
- Number of recordings
- Duration
The task can only be saved if the account balance is sufficient.
2.3 AI QM Task Details

| Field Name | Description |
|---|---|
| Recording Time | Timestamp of the recording |
| Agent Account | Corresponding agent account |
| Agent Group | Selected agent group |
| Call ID | Unique call identifier |
| Status |
Waiting: QM has not started In Progress: QM is ongoing Completed: QM finished Failed: QM failed (only when LLM recognition errors occur) |
| AI Score | Initial AI-generated score |
| AI Comment | Hover to view full comment |
| Review Score | Editable once |
| Reviewer | Agent account who modified the score |
| Recording | Supports playback, volume control, download, playback speed |
| Action | “Retry” available only for failed records |
2.4 AI QM Reports

Supported Filters:
- Agent Group
- Agent Email / Account / Nickname
- Time Range (based on recording time, max 1 month)
Access Control:
- Admins can view all reports under the tenant
- Other roles can only view their own reports
| Field Name | Description |
|---|---|
| Agent Name | Agent name |
| Agent Group | Agent group |
| Evaluation Count | Number of evaluated recordings |
| Average Score | Average score of all recordings |
| Details |
Click to view detailed records
|
2.5 AI QM Consumption Records

Supports filtering of consumption records:
| Field | Description |
|---|---|
| Time Range | Based on QM task start time |
| Consumption Item | Supports selection of AI QM |
| Tenant | Defaults to current tenant (not editable) |
| QM Task Name | Supports searching by specific task name |