Sentiment Analysis
Analyze caller sentiment from transcripts using openai@v2 with any OpenAI-compatible LLM. Detect frustration, satisfaction, and urgency with custom prompts. Results delivered as structured JSON to your CRM or analytics platform via fetch@v2.
Overview
Sentiment Analysis uses openai@v2 to score caller emotion and satisfaction from call transcripts. Unlike rule-based systems, LLM-powered sentiment analysis understands context, sarcasm, and nuance — scoring on your custom scale.
How it works
Call transcripts are captured via startRecording@v1 with automatic transcription. In the hangup branch, the full transcript is sent to your LLM with a sentiment analysis prompt. You define the scoring dimensions: overall sentiment (positive/neutral/negative), frustration level, urgency, satisfaction — any axis relevant to your business.
Custom scoring
Define your own sentiment taxonomy via the prompt. Score on a 1-10 scale, categorical labels, or multi-dimensional axes. Ask the LLM to explain its reasoning. Use response_format: { type: json_object } to enforce structured output.
Integration
Sentiment scores are pushed to your CRM, quality management system, or analytics platform via fetch@v2. Trigger alerts for negative sentiment via email@v1 or sms@v1. Build dashboards by aggregating scores over time.
Others features
We own our network
Global coverage
Order numbers in just one click in over 220 countries, including premium, vanity or toll-numbers.
Encrypted and secure
We offer state-of-the-art data encryption with HTTPS, SIP TLS & SRTP.
Reliable network
With multiple data centers and servers around the world, we offer a robust service you can rely on.
Registered carrier
We operate our own network for better performance, localization and support.
