How does automatic transcription work?

Insights AI transcribes every recorded call automatically. The transcript includes speaker separation: each utterance is attributed to either the agent or the customer, so the full conversation is easy to read and search without replaying the audio.

Transcription is triggered as soon as a call recording is available and requires no manual action.

What sentiment labels does Insights AI use?

Each call is assigned one of three overall sentiment labels: Positive, Neutral, or Negative. The AI also produces numeric scores, both an average score and a net score, giving you a precise measure alongside the label.

These labels and scores are stored as CRM fields and can be used to filter and prioritize call reviews.

What is sentiment evolution?

Beyond the overall label, Insights AI tracks how sentiment changes across the call. The evolution is classified as Improving, Stable, or Deteriorating, and a per-segment sentiment timeline is also available.

This is useful for identifying the point in a call where a customer became frustrated, or conversely where an issue was successfully resolved.

What speech KPIs does Insights AI compute?

In addition to AI analysis, CX-Engine computes a set of speech metrics for every call:

• Speech ratios: the share of speaking time attributed to the agent and to the customer
• Speech rates: words per minute for both the agent and the customer
• Monologue durations: the length of the longest uninterrupted speech segment for each party
• Question ratios: the proportion of utterances that were questions, for both the agent and the customer
• Rephrasings count: how many times the agent rephrased the customer's request

These metrics help identify agents who talk too much, speak too fast, or fail to ask clarifying questions.

What is First Call Resolution (FCR)?

First Call Resolution (FCR) measures whether a customer's issue was resolved on the first contact, without requiring a follow-up call. Insights AI evaluates FCR automatically for each call and stores the result as a boolean field.

FCR is a key quality indicator for support and service teams. Tracking it at scale, across every call, gives managers an accurate and objective measure of resolution effectiveness.