🏟️ Smart Stadium Concierge - FIFA World Cup 2026

An intelligent, AI-powered stadium concierge system designed to enhance the fan experience and streamline stadium operations during the FIFA World Cup 2026. Built for the Hack2skill and Google for Developers Hackathon.

📖 Project Overview

The Smart Stadium Concierge is designed to solve the massive logistical and experiential challenges of hosting global mega-events like the FIFA World Cup 2026. Managing millions of international fans requires real-time, context-aware, and multilingual support that static maps and basic apps simply cannot provide.

By leveraging cutting-edge Generative AI and spatial data retrieval, this platform acts as a hyper-personalized digital assistant. Whether a fan needs to find the nearest wheelchair-accessible restroom, locate specific dietary concessions, or get real-time gate updates, the AI synthesizes complex stadium data into natural, instant answers in any language.

Simultaneously, the platform features a dedicated Ops Dashboard for stadium staff, translating fan queries and congestion data into actionable operational insights.

✨ Features

• 🗣️ AI Fan Concierge: A multimodal conversational interface for fans to ask questions about navigation, concessions, and schedules.
• 🌐 Multilingual Support: Automatically detects and responds in the fan's native language, breaking down language barriers.
• ♿ Accessibility First: Provides wheelchair-friendly routes avoiding stairs.
• 👨‍💼 Ops Dashboard (Staff Portal): Real-time insights and issue tracking for stadium staff.
• 🧠 Generative AI Integration: Powered by Google Gemini through the Vercel AI SDK for intelligent natural language processing.
• 🗄️ Vector Database (Supabase): Uses RAG (Retrieval-Augmented Generation) to ground the AI in actual stadium spatial data and policies.

🧠 Architecture & Generative AI (RAG)

To provide highly accurate, hallucination-free responses, this application implements an advanced Retrieval-Augmented Generation (RAG) pipeline:

1. Knowledge Base & Vector Embeddings:

   • Massive amounts of unstructured stadium data (schedules, concession menus, gate locations, accessibility paths, and security policies) are converted into high-dimensional vector embeddings using Google's embedding models.
   • These embeddings are stored in Supabase, utilizing the pgvector PostgreSQL extension for blazing-fast semantic search.

2. Semantic Search & Context Retrieval:

   • When a fan asks a complex query (e.g., "Where is the nearest halal food to block 104 that is wheelchair accessible?"), the query is vectorized in real-time.
   • pgvector performs an Approximate Nearest Neighbor (ANN) search to retrieve the most contextually relevant data chunks from the database.

3. Google Gemini Integration (Vercel AI SDK):

   • We utilize the Vercel AI SDK to seamlessly stream the retrieved context alongside the user's prompt directly to the Google Gemini language model.
   • Gemini acts as the reasoning engine: synthesizing the raw spatial data into a natural, conversational response, and automatically translating it into the user's native language.

4. Interactive 3D Spatial Awareness:

   • Integrated with React Three Fiber, the AI doesn't just return text—it provides physical context, grounding the RAG results into a tangible 3D environment for the fan.

🛠️ Tech Stack

• Framework: Next.js (React 19)
• AI: Vercel AI SDK & Google Gemini Models
• Infrastructure: Google Cloud Run (Dockerized Serverless Hosting)
• Database: Supabase (pgvector for semantic search)
• Styling: Tailwind CSS & Framer Motion
• 3D Elements: React Three Fiber

🚀 Getting Started

1. Clone the repository
2. Install dependencies:

   npm install

3. Set up environment variables in .env.local (Google AI API Key, Supabase credentials).
4. Run the development server:

   npm run dev

☁️ Google Cloud Platform Deployment

This project is optimized for deployment on Google Cloud Run, showcasing an enterprise-grade, serverless container architecture.

1. Authenticate with Google Cloud:

   gcloud auth login
   gcloud config set project YOUR_PROJECT_ID

2. Build and deploy to Cloud Run:

   gcloud run deploy smart-stadium \
     --source . \
     --region us-central1 \
     --allow-unauthenticated \
     --set-env-vars="GOOGLE_AI_API_KEY=your_key,NEXT_PUBLIC_SUPABASE_URL=your_url,NEXT_PUBLIC_SUPABASE_ANON_KEY=your_key"

👨‍💻 Developed By @idevakk

Built during the Generative AI Hackathon by Google for Developers & Hack2skill.