Typical Scenario: Configuring AI Inspection Tasks
AI inspection refers to the process in which the AICC automatically rates agent call records based on specified rules such as wording, pause, and silence during a call between an agent and a customer. This reduces the inspection workload and labor cost in the call center. In addition, the AICC supports the second inspection by an inspector to supplement the AI inspection result, which avoids misjudgment.
Prerequisites
- You have set positions for employees involved in AI inspection on the
The titles and corresponding operations are as follows:
- Inspection Supervisor: allocates an agent appeal application to an inspector.
- Inspector: performs the second inspection based on the AI inspection result.
- Agent Director: reviews the appeal application submitted by an agent.
- Agents: provides call records for inspection and files possible appeals.
You have synchronized the preceding personnel titles to the AI inspection engine (currently, a third-party AI inspection engine).
By default, the AICC synchronizes information to the AI inspection engine every 10 minutes. If the information modified in the AICC is required, wait for 10 minutes before performing operations.
page.
- You have configured
Context
After the tenant space is enabled, the administrator needs to configure the following items for the AICC to execute AI inspection tasks.
Scenario Description
The following AI inspection scenarios and sample inspection plans are specified in a call center:
- Points are deducted if an agent does not begin with a self-introduction.
- Points are deducted if agent statements contain abusive words during a call.
- Points are deducted if an agent pauses for more than 5 seconds during a call without telling a customer "Please wait".
- Points are deducted if an agent interrupts a customer during a call.
- All call records are inspected intelligently and 5% of them are inspected manually.
- The rating items for manual sample inspection are the same as those for manual post-event inspection.
Procedure
- Configure semantics and keywords.
- Choose .
- Add the following tags by referring to Managing Sentences.
Table 1 Configuring semantics and keywords Tag Name
Tag Type
Description
Self-introduction at the beginning
Semantics
Hello, this is agent ABC of the XXX call center.
Sensitive words
Keyword
Use commas (,) to separate multiple sensitive words.
Waiting prompt
Semantics
Please wait.
Please wait a moment.
Please hold on.
- Configure sentences.
- Choose .
- Click Create Sentence in the upper right corner to add a sentence for each of the three tags in the previous step. For details, see Managing Sentences.
- Click Training Quality Inspection Model in the upper right corner to train semantics, keywords, and related configurations.
- Configuring inspection rules.
- Add a dialog rule by referring to Configuring Dialog Rules.
Based on the scenario, you can add two dialog rules. One is used to check whether an agent begins with a self-introduction, and the other is used to check whether sensitive words are contained in a call.
The following uses the rule Introduction at the call beginning as an example.
During the configuration, pay attention to the following points:
- Scoring Method and Implementation Mode: used to configure whether the rule is mandatory for compliance and whether the score is deducted when the rule logic is met.
For example, if Scoring Method and Implementation Mode are set to Deducted point and This rule is mandatory for compliance, respectively, the rule is mandatory for compliance and no point is added or deducted when its logic is met. When its logic is not met, points are deducted. Similarly, if Scoring Method and Implementation Mode are set to Added point and This rule is forbidden for compliance., respectively, the rule is forbidden for compliance and no point is added or deducted when its logic is not met. When its logic is met, points are added.
- Number of Matched Dialog Logic: The current rule is met only when a specified number of logics are successfully matched in the rule. Occurrence in the Dialog Logic area: The logic is met only when it occurs for the number of specified times.
- Configure the optional items properly, including Description and Trigger Condition. If these items are not configured, the inspection rule cannot be associated in subsequent configuration due to incompleteness.
- Scoring Method and Implementation Mode: used to configure whether the rule is mandatory for compliance and whether the score is deducted when the rule logic is met.
- Add a silence rule by referring to Configuring Silence Rules, as shown in the following figure.
- Add a dialog rule by referring to Configuring Dialog Rules.
- Configure an inspection rating rule. For details, see Configuring Inspection Rating Rules.
- Configure a business scenario.
- For details, see Configuring Business Scenarios.
In the preceding configuration, Inspection Rule for Sensitive Word specifies whether to display sensitive word violations in inspection reports. Currently, the AICC does not rate sensitive words. For details about how to configure sensitive words, see Managing Sentences.
- On the Business Scenario page, click Enable next to the new business scenario.
- For details, see Configuring Business Scenarios.
- Configure a sample inspection plan. For details, see Configuring Sample Inspection Plans.
If no inspector or agent is available, check whether the corresponding titles in Prerequisites have been added for the employees.
Follow-up Procedure
After operations in 5 are complete, the administrator can view the AI inspection result by referring to Viewing AI Inspection Results.
After the operation in 6 is complete, the inspector can execute the inspection task by referring to Creating Sample Inspection Tasks.
Feedback
Was this page helpful?
Provide feedbackThank you very much for your feedback. We will continue working to improve the documentation.See the reply and handling status in My Cloud VOC.
For any further questions, feel free to contact us through the chatbot.
Chatbot