Automated Gmail Routing, Priority Qualification & Owner Assignment

Context

Poly LAB is a leading online school for 3D modeling and animation. Due to rapid scaling and a daily increase in inbound requests (tickets), there was an urgent need to optimize customer support operations. The existing ticket handling process required full automation for classification, precise task distribution within the support team, and strict monitoring of service-level agreements (SLA).

Problem

• High Time Consumption:

The support team spent excessive resources on routine reading, manual classification, and sorting of incoming messages. Analyzing and routing a single ticket took approximately 30 minutes.

• Reputational and Financial Risks:

Due to the chaotic flow of requests and lack of prioritization, critical inquiries and warm leads were getting lost among general questions and spam. This resulted in the loss of potential clients and decreased brand loyalty.

• Lack of Automated Limit Control:

Response time tracking was handled manually, making it impossible to proactively identify overdue tickets or for management to react promptly.

Solution

Smart Routing AI Agent:

An autonomous AI agent that monitors incoming communication channels every minute and instantly analyzes tickets using an LLM. The agent implements Chain of Thought logic, evaluating urgency, business impact, and client context.

• Automatically assigns a category (support, billing, tech, complaint, spam) and priority (P1–P4).
• Calculates an exact response deadline according to the SLA.
• Assigns a dedicated ticket owner.
• Analyzes customer sentiment, creates a concise text summary, and logs the reasoning behind its decision.
• Synchronizes all data with a central Google Sheet and instantly sends a structured notification to the responsible employee via Slack, complete with a direct link to the Gmail thread.

SLA Control & Escalation Watcher:

An autonomous trigger that monitors the database of open tickets every 5 minutes and compares the current time with the calculated SLA deadline. In the event of a delay, the system initiates a 3-level escalation logic: