AI Recommendation Chatbot

Background

An antivirus platform wanted to help customers find the right product for their needs. The existing product comparison page was static and overwhelming, leading to high bounce rates and support tickets from confused users.

Goal

Build an AI-powered chatbot that understands customer needs through natural conversation and recommends the most suitable antivirus products using RAG (Retrieval-Augmented Generation) with the company's product catalog.

What I built

• Designed and implemented a RAG pipeline using Pinecone as the vector database to store and retrieve product information based on semantic similarity
• Integrated the OpenAI API for natural language understanding and response generation, with carefully crafted prompts to ensure accurate product recommendations
• Built a Vue.js chat interface with real-time streaming responses, conversation history, and product comparison cards
• Developed the Node.js backend to orchestrate the RAG pipeline, manage conversation state, and handle the embedding and retrieval workflow

Outcome

The chatbot successfully guided users to appropriate products through natural conversation. It reduced product-related support tickets and provided a more engaging alternative to the static comparison page, helping users make confident purchasing decisions.