Project Overview
This project showcases the integration of a cutting-edge AI customer support agent into a legacy Shopify/custom MERN storefront. Utilizing Retrieval-Augmented Generation (RAG), the bot retrieves product availability, store policies, and user purchase histories, answering natural language queries and processing refunds or order cancellations with minimal human intervention. Data security is strictly preserved, and conversational state is synced using Redis caching.
Key Features & Scope
Conversational shopping assistant showing product listings directly in chat
Semantic search indexing entire product catalogs for fast discovery
Secure integration with order shipping endpoints to provide tracking updates
Automatic escalations to live human support agents upon sentiment detection
System Architecture
Built with Node.js and Express backend wrapping OpenAI API, managing context size via LangChain buffers. Conversational history and vector embeddings are stored in Pinecone and MongoDB.
Biggest Challenge & Resolution
The Challenge
The online storefront struggled with a high volume of repetitive support queries regarding order tracking and product recommendations, leading to slow response times and cart abandonment.
The Resolution
Built a LangChain-powered RAG chatbot utilizing OpenAI's GPT-4 and Pinecone vector database. It securely accesses live Shopify order tracking APIs and semantically searches the catalog, resolving 70% of support tickets instantly and boosting conversion rates by 18%.