AI call analysis and analytics generation system

Developed an automated AI system for processing and analyzing phone calls with subsequent analytics and reporting.

The system allows for automatic processing of audio calls, determining sentiment, category, and obtaining statistics in real-time.

How the system works:
the system regularly checks for new calls
audio is automatically sent for transcription (AssemblyAI)
the received text is analyzed through AI
sentiment is determined: positive / negative / neutral
the category of the inquiry is determined
data is stored in a database and Google Sheets

AI functionality:
analysis of the conversation content
classification of calls by sentiment
determination of the inquiry category
AI agent for obtaining statistics on request
working through tool calling (obtaining accurate data without fabrication)

Analytics and automation:
counting the total number of calls
distribution by sentiment
statistics by categories
calculation of average call duration
ability to obtain reports for a period

Additionally:
automatic calculation of processing costs (OpenAI + AssemblyAI)
error handling system (logging in Google Sheets)
monitoring processing statuses (queue → processing → done)

Result:
fully automated call processing
obtaining analytics without human involvement
reduction of time for conversation analysis
ability to quickly assess the quality of communication with clients
control of expenses on AI services

Technologies:

n8n, Webhooks, AI Agents, OpenAI, AssemblyAI, Google Sheets, API integrations