AI Customer Ticket Routing Automation is an academic prototype built using n8n and Large Language Models (LLMs) to automate customer ticket processing.
The workflow receives customer messages, extracts relevant information, enriches customer context, analyzes message content using AI, and routes tickets to the appropriate handling path.
- Automate customer message processing.
- Classify customer sentiment and inquiry categories.
- Assess ticket priority using AI.
- Retrieve customer information to enrich ticket context.
- Route tickets to relevant teams.
- Store ticket information and processing results for tracking.
The automation consists of the following stages:
- Input: Receives customer messages through Telegram and Webhook triggers.
- Data Normalization: Extracts and structures incoming message information into a consistent format.
- Customer Context Enrichment: Looks up customer records using Google Sheets.
- AI Analysis: Uses an LLM through OpenRouter to analyze customer messages and generate structured results.
- Decision & Routing: Directs tickets according to their category and analysis results.
- Notifications: Sends notifications through Telegram or email, depending on the workflow path.
- Data Persistence: Stores ticket details and processing results in Google Sheets and supports ticket status tracking.
- Customer message analysis.
- Sentiment classification.
- Inquiry category identification.
- AI-assisted priority assessment.
- Customer information lookup.
- Automated ticket routing.
- Notification automation.
- Ticket logging and status tracking.
- n8n – Workflow automation.
- OpenRouter – LLM integration.
- Large Language Models (LLMs) – Message understanding and classification.
- Google Sheets – Customer lookup and ticket records.
- Telegram Bot API – Message input and notifications.
- Gmail – Email notifications.
This project was developed as an academic automation prototype to demonstrate the integration of AI analysis with workflow automation.
It was tested as a prototype and is not presented as a production deployment or as a system connected to a live company's customer database.
Any publicly shared workflow configuration or screenshots should exclude API keys, credentials, private identifiers, customer information, and other sensitive data.
Workflow screenshots and sanitized configuration examples can be added to this repository to illustrate the automation process.