Automatic analysis
After each call ends, AssistPulse automatically analyzes the transcript using AI to generate:- Summary — a concise overview of what was discussed and any actions taken
- Sentiment — the overall emotional tone of the caller (positive, neutral, negative)
Summaries
Call summaries capture the key points of a conversation in a few sentences. They typically include:- The caller’s main question or request
- How the agent responded
- Whether the issue was resolved
- Any follow-up actions (transfers, callbacks promised)
Sentiment analysis
Sentiment is classified into three categories:| Sentiment | Meaning |
|---|---|
| 🟢 Positive | Caller was satisfied, friendly, or happy with the interaction |
| 🟡 Neutral | Standard informational call, no strong emotion |
| 🔴 Negative | Caller was frustrated, dissatisfied, or upset |
Using sentiment data
- Monitor quality — a high ratio of negative sentiment calls may indicate issues with your agent’s instructions or knowledge base
- Prioritize follow-ups — review negative sentiment calls first to identify problems
- Track trends — watch sentiment over time to measure improvement
Sentiment analysis requires transcription to be enabled. Calls without transcripts don’t receive sentiment scores.