Let's connect.

k.allenpace@uga.edu

Incoming support emails are triaged and responded to without human intervention. The knowledge base ensures replies are grounded in accurate information rather than model assumptions. Built and demonstrated as a proof of concept.

The Outcome

An automated support workflow that monitors a Gmail inbox, uses a text classifier to distinguish customer support emails from everything else, and routes support emails to an AI Agent backed by a Pinecone knowledge base. The agent retrieves relevant content and drafts a reply grounded in the knowledge base. Non-support emails are ignored. Support threads are labeled and replied to automatically.

What I Built

Support email volume does not scale proportionally with headcount. As a business grows, manual triage and response becomes a bottleneck. Every email that requires a human to read, classify, and draft a reply is time that could be spent elsewhere.

The Problem

Customer Support Workflow