GenAI Email Processor

Project Overview

The GenAI Email Processor is an AI-powered system designed to automate email classification, order processing, and customer inquiries using Large Language Models (LLMs). It processes emails from a given dataset, categorizes them, checks product availability, generates order confirmations, and responds to product-related questions—all while dynamically updating stock levels.

Key Features

Email Classification – Automatically categorizes emails as either "product inquiry" or "order request" using LLM-based intent detection.

Order Processing & Stock Management –

• Verifies product availability in real time.
• Updates stock levels based on order fulfillment.
• Generates order statuses ("created" or "out of stock").

Automated Response Generation –

• Sends personalized, production-ready emails for:
  • Successful orders (with product details).
  • Out-of-stock items (with alternatives or restock options).
• Handles product inquiries by retrieving relevant details from a large catalog (100k+ products) without exceeding token limits.

Scalable AI Techniques –

• Uses Retrieval-Augmented Generation (RAG) and vector stores for efficient data retrieval.
• Optimized for GPT-4o (OpenAI API) to balance cost and performance.

Technical Implementation

• Built with Python using LLM frameworks (e.g., LangChain).
• Processes structured inputs from Google Spreadsheets (products & emails).
• Outputs organized results in a multi-sheet spreadsheet for easy tracking.

Use Case

This system is ideal for e-commerce, customer support automation, and inventory management, reducing manual workload while ensuring accurate, AI-driven responses.

---

How It Works

1. Inputs:

   • Product Catalog (ID, name, category, stock, description, season).
   • Email Dataset (ID, subject, body).

2. Outputs:

   • Email Classification Sheet (email ID, category).
   • Order Status Sheet (email ID, product ID, quantity, status).
   • Order Response Sheet (email ID, response).
   • Inquiry Response Sheet (email ID, response).

3. AI-Powered Logic:

   • LLM-driven classification & response generation.
   • Smart inventory updates post-order processing.
   • Efficient product search without full catalog injection.