Skip to content

At a glance

This page provides a quick introduction to the AI-powered analytics in the MiaRec platform.

The MiaRec platform automatically analyzes every call to provide:

  • Transcription.
  • Call Summary.
  • Sentiment score.
  • Topic analysis
  • AI Insights (key facts, next actions, reason for the call, etc.)
  • Auto QA (AI-powered automatic agent evaluation)

Info

Some of the features may not be available to you due to the permissions and licensing set up by the administrator.

Transcription

Every call is being transcribed automatically in the MiaRec platform.

Key features of MiaRec Transcription:

  • Automatic. Every call is being transcribe automatically
  • Accurate. MiaRec uses the latest technologies to provide exceptional transcription accuracy, even in noisy environment and accented speech.
  • Speaker separation. See in transcript who said what. In case of phone platform doesn't provide dual-channel audio, the MiaRec platform automatically detects speakers using a Speaker Diarization model.
  • Multilingual. MiaRec can transcribe calls in different languages. And it can even transcribe calls with intermixed languages in the same conversation.
  • Full-text search. You can search through all your calls using a text.
  • Post-call. Transcription is provided after call is completed.
  • Inline comments. A reviewer can add inline comments in the transcript as a feedback to an agent.

View transcript

Call Summary

The MiaRec platform automatically creates a short summary of each call.

Key features of MiaRec Call Summary:

  • Automatic. Call summary is created automatically by AI.
  • Customizable. You can use the AI Prompt Designer to customize the call summary structure by instructing AI to include certain things into a call summary
  • Editable. The authorized users can edit the call summary if necessary.

View call summary

Sentiment analysis

Sentiment analysis detects the underlying emotion in customer calls and classifies it as positive, negative or neutral. Every call is assigned a sentiment score value (in a range from -100 to +100) and a short explanation of the score is provided as an evidence. In case of a negative score, the transcript highlights the phrases that lead to that score.

Key features of MiaRec Sentiment analysis:

  • Highly accurate. MiaRec uses Large Language Models to analyze a whole conversation to provide an accurate sentiment score.
  • Separate agent and customer score. A sentiment score is provided separately for both sides of the conversation, agent and customer
  • Explanation of a score. Each score is provided with an explanation why it a call was assigned such a score.
  • Highlight negative phrases. The most negative phrases that lead to a negative score are highlighted in the transcript
  • Customizable. Using the AI Prompt Designer, you can instruct AI scoring engine to score negatively calls that match certain criteria for you, like problem was not resolved on the first call, etc.

View sentiment score

Sentiment highlighted phrases in transcript

A high-level view of sentiment scores is provided in the Dashboard, where you can narrow down your search to group, user, sentiment, etc.

Sentiment dashboard

A sentiment heatmap allows you to quickly locate users with high volume of negative calls on a timeline.

Sentiment heatmap

Topic analysis

MiaRec analyzes call transcripts to identify topics of conversation, like "Service cancellation", "Escalation to manager", "Credit card declined", "Issue with phone", "Reschedule appointment", etc.

Key features of MiaRec Topic analysis

  • Multiple topics per call. Multiple topics can be assigned to the same call
  • Quick search and filter by topic.
  • Trend analysis.

Topic analysis

Topics dashboard

AI Insights

MiaRec AI Insights module allows you to analyze every calls and extract any insights from the conversation.

Example of AI insights:

  • Reason for the call. AI Assistant can choose from one of the pre-defined options, or it can optionally create new options if the call doesn't fall into one of existing categories.
  • Product information.
  • Action items. What the agent must do after the call.
  • Knowledge gap. Identify the questions that the agent was struggling to answer with certainty.

The key features of MiaRec AI Insights:

  • Customizable. You can customize AI Prompt to instruct AI to process the data the way you want.
  • Prompt designer. You can use the MiaRec AI Playground to test different prompts on your own data.

AI Insights

Report Reason for the Call

Auto QA

MiaRec uses AI to automatically score calls using agent evaluation form.

Key features of the MiaRec Auto QA:

  • Easy and intuitive configuration. You can specify the scoring criteria in plain language, like "Did the agent introduce themselves properly?". Optionally, you can provide some clarification in a description, for example, explain that "appropriately" means thanking the caller for calling, asking how they are doing, offering a help, etc.
  • Evidence for each answer. AI Assistance provides an evidence of its answer for each question. It explains why it choose certain answer.
  • Override score if necessary. A reviewer can manually override the AI answer to recalculate the score.

Auto QA report

For the more detailed description of each of these features, check the corresponding sections in this document. `