Mall-Mind is a mobile-first Progressive Web Application (PWA) designed to transform traditional mall experiences using Agentic AI.
It brings e-commerce–level personalization, intelligence, and automation into physical retail environments.
The system uses multiple intelligent agents to understand user intent, check simulated real-time inventory, generate optimized indoor routes, and provide dynamic shopping assistance — all without requiring app installation.
With the rapid growth of platforms like Amazon and Flipkart, physical malls struggle with personalization and digital intelligence. Shoppers often feel overwhelmed by numerous stores and lack real-time visibility into inventory, discounts, and navigation. Mall-Mind bridges this gap by combining AI-driven personalization with real-world shopping experiences.
MallMind
│
├── backend
│ ├── agents
│ │ ├── stylistAgent.js
│ │ └── inventoryAgent.js
│ │
│ ├── graphql
│ │ ├── schema.js
│ │ └── resolvers.js
│ │
│ ├── navigation
│ │ └── mallGraph.js
│ │
│ ├── models
│ │ └── User.js
│ │
│ ├── middleware
│ │ ├── auth.js
│ │ └── rateLimiter.js
│ │
│ ├── tests
│ │ └── api.test.js
│ │
│ ├── server.js
│ └── package.json
│
├── frontend
│ ├── pages
│ │ ├── index.js
│ │ └── chat.js
│ │
│ ├── components
│ │ ├── ChatBox.js
│ │ └── StoreCard.js
│ │
│ └── package.json
│
├── docs
│ ├── synopsis.pdf
│ └── architecture.png
│
├── .github
│ └── workflows
│ └── ci.yml
│
├── README.md
└── docker-compose.yml
Physical malls currently:
- Lack personalized store recommendations
- Do not provide real-time inventory visibility
- Offer static navigation maps
- Cannot optimize walking routes
- Do not leverage session-level visitor data
To design and develop a scalable Progressive Web Application (PWA) that:
- Understands shopping intent using natural language
- Recommends relevant stores based on preferences
- Checks simulated real-time inventory across brands
- Generates optimized indoor walking routes
- Enables AI-powered virtual try-on experiences
- Supports dynamic promotional logic
Mall-Mind follows a modular layered architecture:
- Built using React + Next.js
- Mobile-first PWA (no installation required)
- Interactive UI for shopping input and navigation
- Developed using Node.js + GraphQL
- Handles session management and API orchestration
- Connects AI agents with store data
- PostgreSQL for structured session data
- MongoDB for flexible inventory and interaction data
- Powered by OpenAI Assistants API
- Coordinates autonomous agents (Stylist, Scout, Navigator, Negotiator)
- Integrated with Mapbox Indoor SDK
- Provides optimized in-mall navigation
Visitors scan a QR code at the mall entrance to instantly start a session.
Users describe their needs conversationally:
"I need a blue suit under $300 and coffee."
The AI extracts product type, color, budget, and preferences.
- Queries simulated store APIs
- Matches availability, size, price, and color
- Recommends relevant stores only
- Understands style, occasion, and budget
- Provides tailored product recommendations
- Interactive mall map
- Optimized walking routes
- Dynamic rerouting when required
- Users upload a selfie
- AI simulates clothing fit
- Improves purchase confidence
| Stakeholder | Value Delivered |
|---|---|
| Shoppers | Faster, personalized, stress-free shopping |
| Store Managers | Increased engagement and conversion insights |
| Mall Admin | Visitor analytics and operational efficiency |
- JWT-based authentication
- HTTPS encrypted communication
- Temporary anonymized sessions
- No permanent storage of personal images
- Automatic session expiry
- ≥30% reduction in shopping time
- ≥40% improvement in store relevance
- ≥20% increase in conversion via dynamic offers
- ≥95% indoor navigation accuracy
# Clone repository
git clone https://github.com/your-username/mall-mind.git
# Backend setup
cd backend
npm install
npm start
# Frontend setup
cd frontend
npm install
npm run dev