This guide provides supervisors with a comprehensive overview of Ascend Teams AI Supervisor Assist and AI Agent Evaluator. Our goal is to equip you with the knowledge to leverage these features effectively, ensuring you can maximize their benefits for real-time coaching, agent support, and overall performance improvement.
What is AI Supervisor Assist?
AI Supervisor Assist gives Contact Center supervisors real-time tools to coach more effectively, recognize strong performance as it’s happening, and step in when calls need support, all while complementing your existing post-call review process. With features such as live sentiment monitoring, call transcription, AI-powered summaries, queue insights, and in-call feedback tools (such as whisper and UC Chat), it helps supervisors take action during live interactions and reinforce learnings afterward.
Key capabilities include:
- Real-time AI insights on active calls
- Searchable live call transcripts
- Live call topics with sentiment
- Evolving sentiment monitoring
- Real-time summaries for intervention decisions
Real-Time Queue Insights
Supervisors gain crucial real-time queue insights from the Real-Time Monitoring page. Key metrics such as hold times, call volumes, and call durations instantly highlight staffing gaps, workflow issues, or coaching needs.
What to Look For
- Spikes in average hold time or call duration
- Agents consistently handling longer calls
- Sudden shifts in queue volume
- Queues with little activity during peak hours
How to Act on It
- Shift agents between queues
- Investigate longer calls
- Monitor underused queues
- Use queue-level data to spot coaching trends
- Pair queue data with sentiment and summary trends
Live Sentiment Analysis
Live Sentiment Analysis gives supervisors real-time insights into the call’s emotional tone. It uses AI to analyze the words and tone customers use, updating sentiment throughout the conversation.
Sentiment Colors
- Green: Positive, cooperative language
- Red: Frustration, confusion, or negative sentiment
- Grey: Neutral or unclear language
Best Practices
- Check transcripts when sentiment changes
- Use whisper or chat when red sentiment persists
- Save red-to-green recovery calls as training examples
- Don’t only focus on negative calls; recovery calls are valuable coaching opportunities
Live Transcription
Live transcription provides a real-time, searchable record of every call, clearly distinguishing between the agent and customer. Every statement is tagged with sentiment and can be searched by keyword.
What to Look For
Support Environments
- Repeated customer questions
- Terms such as “refund,” “cancellation,” or “supervisor”
- Gaps in agent responses
- Objections in customer language
Sales Environments
- Missed buying signals
- Leading questions from agents
- Sentiment shifts in context
General
- Unusual call flow
- Searchable coaching moments
AI-Powered Call Summaries
Mid-call summaries provide supervisors with real-time insight into ongoing conversations without needing to join every call. Calls are automatically tagged with topics and sentiment.
Best Practices
- Prioritize calls with multiple negative topics
- Identify successful de-escalation examples
- Use topic-level sentiment for targeted coaching
Chat, Whisper, and Barge
AI Supervisor Assist includes:
- Monitor: Listen silently
- Whisper: Coach agents privately
- Barge: Join customer conversations
- UC Chat: Private messaging to agents during calls
Coaching Tips
- Use Whisper for quick corrections
- Use Chat for detailed references and links
- Avoid micromanagement
- Reinforce positive behaviors as often as corrective feedback
AI Agent Evaluator
AI Agent Evaluator helps supervisors review and score calls more efficiently through keyword-based or manual call selection. Evaluations include:
- Full transcripts
- Sentiment tagging
- Keyword highlights
- Evolving sentiment analysis
- Talk-to-listen ratio metrics
- Words-per-minute measurements
Evaluation Features
Evaluations Dashboard
- Agent evaluation list
- Evaluation search
- Inbound/outbound/external evaluation types
- Evaluation status tracking
- Collaboration and acknowledgment tracking
- Custom evaluation templates
Grading Templates
Predefined and customizable evaluation criteria with:
- Yes/No questions
- Multiple choice questions
- Rating systems
Interaction Analytics
- Overall sentiment analysis
- Agent sentiment
- Customer sentiment
- Talk/Listen Ratio
- Words per Minute (WPM)
- Keyword Finder
Coaching Recommendations
Focus on:
- Calls containing “cancel,” “escalate,” or “refund”
- Repeated evaluation deficiencies
- Talk speeds above 160 WPM
- Heavily skewed talk/listen ratios such as 80/20
Best Practices
Do’s
- Monitor real-time sentiment cues
- Use whisper and chat for immediate coaching
- Validate AI-generated information
- Use queue data for staffing decisions
- Provide constructive feedback based on AI insights
Don’ts
- Rely solely on AI sentiment
- Undermine agent trust through excessive monitoring
- Ignore agent requests for assistance
- Micromanage agents
- Assume AI summaries are always 100% accurate
