Obrabot: Say Goodbye to Manual Call Reviews

Obrabot is an internal service designed to lift the burden off Quality Assurance (QA) departments. Instead of listening to every call manually, the system does it in seconds using AI.

🔴 The Problem

In companies with high call volumes, QA managers face:

1. Overload: It’s physically impossible to review 100% of calls.
2. Subjectivity: Different managers might rate the same errors differently.
3. Latency: Quality reports often appear days after the actual conversation.

⚡️ The Solution: AI Segmentation

The system integrates directly with Google Sheets, where links to recordings or transcripts are stored.

1. Automatic Import

The service pulls new rows from spreadsheets, processes audio via Whisper (or takes existing text), and sends it to the analytical core.

2. Smart Classification

AI evaluates the call based on predefined criteria:

• Was the call successful?
• Were objections handled?
• Was there any aggression or inappropriate language?
• Was it an answering machine?

3. Interactive Reports

Results are returned to a user-friendly interface where managers can filter only the “problematic” calls that require human attention.

🛠 Tech Stack

• Next.js: Core interface and API.
• Prisma + PostgreSQL: Storage for report history and metadata.
• OpenAI API: Segmentation and Sentiment Analysis.
• Google Sheets API: Two-way data synchronization.

---

Insights: Using AI in QA allows companies to not just find errors, but to see systemic sales patterns that were previously hidden in thousands of hours of audio.