Agentic AI Stack for Intelligent Customer Communication in Ecommerce
P30-T2-Agentic AI Stack
CS 4850-01 Fall Semester 2025
Professor Sharon Perry
Ecommerce websites handle millions of requests in short amounts of time. Agentic systems have shown promise in automating e-commerce web queries and handling complex customer requests. We wanted to build one. Our multi-agent model accepts user queries from a React dashboard, classifies inputs, assigns priorities, distributes work to specialized LLMs, and outputs appropriate responses.
System Capabilities:
- Answer User Questions About Policy and Orders
- Change Account Information
- Process New Orders, Track Old Orders
- Issue Receipts Via Email and Text Message
- Track, Store, and Remember User Conversations
