Skip to content

Repository files navigation

ShopSphere – AI Avatar for E-Commerce CX

🚀 Overview

ShopSphere is a real-time AI Avatar solution built for modern e-commerce experiences. It enables users to interact using text or voice and receive intelligent responses powered by enterprise knowledge, personalization, and AI-driven recommendations.


🔍 Key Features

  • 🤖 AI-powered conversational agent
  • 🧠 RAG-based knowledge retrieval (Azure AI Search)
  • 🎯 Personalized product recommendations
  • 🔄 Conversational correction (user can change decisions anytime)
  • ⚡ Real-time streaming responses (text + voice + avatar)
  • 🛒 Action automation (add to cart, create ticket, etc.)
  • 🧩 Modular multi-agent architecture

🏗️ Architecture Highlights

  • Copilot Studio for routing & orchestration
  • Azure OpenAI for LLM-based planning & response generation
  • Azure AI Search for knowledge retrieval (RAG)
  • Azure Functions for APIs & microservices
  • Azure Cache (Redis) for fast response optimization
  • Azure Cosmos DB for session and state management
  • Azure Speech for voice and avatar interaction

👥 Team Roles

  • Loveleen Batra – AI Core (RAG, OpenAI, Recommendation Engine)
  • Sanket Kumar Mahapatra – Frontend UI + Avatar + Streaming
  • Sunny Raj – Copilot Studio + Routing + State Management
  • Manish Kumar Sahoo – CRM, Actions, Power Automate

🗂️ Project Structure

frontend/ → Angular UI backend-functions/ → Azure Functions APIs ai-core/ → LLM + RAG + Recommendation logic copilot-config/ → Copilot Studio configuration workflows/ → Power Automate flows infra/ → Azure infra setup

⚡ Tech Stack

  • Azure OpenAI
  • Azure AI Search
  • Azure Functions
  • Azure Cache for Redis
  • Azure Cosmos DB
  • Azure Speech Services
  • Angular
  • Copilot Studio

🎯 Hackathon Goal

Deliver a scalable, low-latency AI-powered customer experience platform that improves conversion, reduces support load, and enhances engagement.


💡 Key Improvements

  • Parallel retrieval to reduce latency
  • Streaming pipeline for better UX
  • Dynamic output selection (text/voice/avatar)
  • State-aware conversation handling
  • Confirmation-based action execution

🔮 Future Scope

  • Multilingual support
  • Advanced personalization
  • Human agent handoff
  • AI analytics and insights