Agent jobs are inherently stressful. With advancements in AI, human agents are increasingly managing complex interactions that demand deep knowledge, advanced problem-solving, and refined soft skills. The high cost of hiring, training, and turnover highlights the importance of prioritizing agent well-being. Supporting their engagement and mental health is key to ensuring they consistently deliver exceptional service.

Cisco's AI-powered agent burnout detection provides insights into agents' stress levels in real-time. These insights can be used to implement proactive supportive measures to help agents manage stress, improve engagement, and enhance customer satisfaction. Offering timely wellness breaks, optimizing channel allocation, shift schedules, staffing, and routing strategies are some actions that can be taken to help agents manage stress and deliver positive outcomes for both customers and the business.

Along with burnout detection, we offer timely wellness breaks through the AI Assistant to help agents manage their stress and stay engaged.

How it works

The agent burnout detection model uses Cisco’s proprietary algorithms, which analyze highly specific call metrics and transcripts to identify signs of stress among agents in response to contact center stressors. Since each agent's response to stress is unique, the burnout detection is tailored for every individual agent.

Based on the events from the burnout detection model, well-being breaks are offered to agents through the AI Assistant. These breaks are 1 minute long and provided to agents between their call assignments. A unique system-generated idle code, 'WellbeingBreak,' is used when the breaks are initiated, allowing them to be tracked and reported separately. Agents will be recommended a maximum of one well-being break every 90 minutes and a maximum of three breaks per shift.

Insights are also available through APIs, enabling customers and authorized partners to tailor supportive actions or integrations based on specific agent needs. For more details on accessing the Agent Wellbeing APIs, refer to Agent Wellbeing developer documentation.

This feature is optional and can be activated or deactivated at the discretion of customer administrators for each user, allowing them to control its implementation.

For additional details on how the AI model works, refer to the AI Transparency Technical Note.

Benefits

• Real-time Insights—Monitor agent stress levels to facilitate proactive measures.
• Improved Agent Well-being—Develop strategies to reduce stress and enhance agent performance.
• Enhanced Performance—Boost and sustain efficiency, effectiveness, and overall performance metrics.
• Improved CSAT Scores—Increase customer satisfaction through better agent well-being and performance.
• Reduce Churn—Investing in agent well-being boosts morale, prevents burnout, and helps reduce churn.

Who can use the feature

Agent burnout detection is useful for the following stakeholders:

• Administrators and Partners—Enable and configure the agent burnout detection settings.
• Supervisors—Support and guide agents in managing their stress.
• Agents—Benefit from automated wellness breaks to manage stress levels effectively.
• Developers—Integrate agent burnout detection APIs to analyze stress levels and automate supportive actions.

  For detailed information on the ethical and technical considerations of Agent Burnout Detection, refer to the AI Transparency Technical Note.

Usage guidelines

The agent burnout detection feature should be handled with strict confidentiality and ethical consideration, ensuring that data is solely used to support agent well-being and improve customer satisfaction. The model’s outputs should not be used for assessing an agent’s performance, conducting performance appraisal discussions, penalizing or compensating agents, or informing decisions about an agent’s employment and right to work.

When enabling this feature, transparency is recommended. Inform agents when the feature is enabled and that it is AI-powered.

Prerequisites

• Customers who have purchased the new AI Assistant add-on SKU (A-FLEX-AI-ASST).
• Voice-only interactions—Burnout detection and wellness breaks are supported for agents handling voice interactions. Agents who exclusively manage only digital interactions will not receive wellness breaks.
• English-only interactions—Burnout detection is supported for only English interactions. Do not enable burnout detection and wellness breaks for agents handling non-English languages.
• Stand-alone desktop application—Wellness breaks are supported only for agents using Agent Desktop as a stand-alone application.
• Call recording—Ensure that the call recording is enabled to facilitate post-call processing, which is necessary for effective analysis and support.

Before you begin: Enable call recording to facilitate post-call processing, which is necessary to deliver automated wellness break reminders.

Administrators can enable agent wellbeing feature to activate the agent burnout detection feature either for the entire organization or for specific users. Once activated, the system monitors stress and burnout levels and provides well-timed wellness breaks for agents when needed.

Follow these steps to enable burnout detection and wellness break reminders for agents in the contact center:

1. Sign in to Control Hub and navigate to Services > Contact Center.
2. Go to AI Assistant under Desktop Experience in the navigation pane.
3. Toggle on Detect agent burnout to enable burnout detection for agents.
4. Check Send wellness break reminders to enable wellness breaks for agents.
5. Choose one of the following options:
   • All agents—Select this option to calculate the burnout levels of all agents in the contact center organization.
   • Individual agents—Select specific agents from the list. You can search and filter the agents by name and team.
6. Select the Idle Codes to exclude planned idle time activities from the model's input. The burnout detection model monitors agents' 'idle' state to account for genuine idle time. Exclude any planned activities that do not represent actual idle time, such as team meetings, after-call work, and training.

Now that you have enabled this feature, inform the selected agents that this AI feature is available in their Agent Desktop.

Access the AI Assistant Dashboard in Analyzer to review the Agent Summary and Wellbeing report. This report provides comprehensive insights into the number of wellness breaks provided to agents, along with their workload and performance metrics. It helps evaluate the effectiveness of the wellness feature within your organization.

Additionally, monitor Customer Satisfaction (CSAT) trends through the Auto CSAT table. This enables you to assess how wellness breaks impact agents' wellbeing and performance, which in turn influences customer satisfaction.

For more information, refer to the AI Assistant Dashboard article.

Insights offered by Agent Summary and Wellbeing report

Use the report to track, analyze, and optimize agent wellbeing, performance, and customer satisfaction effectively:

• Wellness Breaks—How many wellness breaks did an agent receive during a specific time period?
• Workload & Performance—What was the agent’s workload and performance during that period, and how have these metrics improved since the wellness feature was enabled?
• Customer Satisfaction—What has been the impact of the wellness feature on customer satisfaction, and how have CSAT scores changed over time?
• Comparative Analysis—How do the Agent Summary metrics compare to those of agents who do not have this feature enabled?

This section answers frequently asked questions about using the AI Assistant for your organization.

1. Will the wellness breaks impact customers' wait time or the workload on other agents?

   Wellness breaks are only a minute long, and the AI Assistant offers a maximum of three breaks during a shift, with only one break allowed in a 90-minute window. This means it will not take away more than three minutes of an agent's working time. These timely micro-breaks have been proven to boost and sustain agents' performance and customer satisfaction.

2. When can you receive burnout scores for agents and wellness breaks after enabling the feature in Control Hub?

   After the feature is enabled for an agent, model training will be initiated. The model is trained using a combination of call data, operational data, and transcripts. To ensure a successful training process, a minimum of 120 voice interactions per agent from the last 30 active days is required. This baseline is critical for establishing a strong starting point for the model's performance. Once model training is completed, burnout predictions will occur for all subsequent voice interactions handled by the agent, and wellness breaks will be offered based on the predicted scores.

3. Can I add or remove agents from this feature?

   Yes, agents can be added or removed from the feature at any time through the Control Hub. Adding new agents will initiate training for them, while removing agents will result in the deletion of their trained models during the next off-hours for the data center.

4. Can the wellness breaks be turned off for unexpected or critical situations?

   Yes, wellness breaks can be enabled or disabled in the Control Hub by selecting or clearing the Send wellness break reminders checkbox under the Agent Wellbeing feature in the AI Assistant.

5. What does it mean if an agent receives wellness breaks or does not receive any wellness breaks?

   Burnout scores for agents are predicted at the end of every interaction. Wellness breaks are offered to agents when their burnout levels cross a specific threshold. If an agent does not receive any wellness breaks, it may indicate that their burnout levels have remained below the threshold, suggesting they are effectively managing their workload. Conversely, if an agent receives more wellness breaks, it indicates higher burnout scores, signaling increased stress levels. Offering wellness breaks at this critical point has been proven effective in helping agents manage stress and regain focus.

6. How often are the Agent Burnout models retrained?

   The Agent Burnout models are retrained every 15 days using data from the preceding 30 days. This regular retraining helps ensure the models remain accurate and free from drift.