An automated system for call transcription, sentiment detection, and analytics built with n8n and LLMs.

Project details

Context:

• Business/industry: Sales / Support / Contact Center
• Current process: Call recordings were stored in Google Sheets as links to audio files. The team manually listened to a portion of the calls, noted issues, and collected basic statistics.

Problem:

• Manual call analysis took too much time and did not scale as call volume increased.
• There was no single place to store transcripts, sentiment, tags, metrics, and reports.
• It was not possible to quickly identify negative calls or build analytics for a specific period without manual work.

Solution

<aside> 🧠

The AI call analytics system built with n8n automatically processes audio, generates transcripts, detects sentiment, extracts insights, and stores the results in a table for reporting and search. Tools used: n8n, LLM, Google Sheets, AI transcription and analysis tools.

</aside>

Solution logic:

1. Audio links are read from Google Sheets.
2. The audio is sent to AssemblyAI for transcription.
3. The transcript is processed by an LLM to analyze sentiment and extract key signals.
4. The results are stored in Data Tables with structured fields for filtering and reporting.