AI system that automatically transcribes, evaluates and escalates problematic customer support calls, with a person checking before any letter to the customer.

Project details

Context:

• Business/industry: Customer Support / Call Center QA
• Current process: Manual call review by QA team — listening to calls, scoring subjectively, no consistent escalation path, no automated client follow-up.

Problem:

• Manual QA review is slow and inconsistent between reviewers.
• Bad calls often go unnoticed until a client complains directly.
• No structured process for turning a bad call into a client-facing apology and resolution.

Solution

<aside> 🧠

A Google Drive folder watch triggers transcription and AI scoring on every new call recording. Calls scoring below threshold automatically escalate to a manager via Telegram with a pre-drafted client email — nothing reaches the client without explicit human approval.

Tools used: n8n, Google Drive, Whisper-compatible transcription, Groq (Llama 3.3), Google Sheets, Telegram, Gmail.

</aside>

Solution logic:

1. Google Drive is polled hourly for new audio files in a designated calls folder.
2. Audio is transcribed (Whisper-compatible output format).
3. An AI model scores the call across 8 QA categories (greeting, active listening, needs discovery, empathy, expertise, solution clarity, next steps, closing), 0-10 each, with a structured summary and feedback.
4. Results are logged to Google Sheets as the single source of truth for every call.