Zivra

Inspiration

We walk into malls and everything looks modern — but behind the scenes, mall operators still use spreadsheets and WhatsApp to manage hundreds of tenants and thousands of daily shoppers. Tenants have no data about their own store performance. Shoppers have no intelligent help finding what they need. We wanted to build the AI layer that malls in India are missing.

What it does

Zivra has three portals:

Shopper — AI Concierge (Gemini-powered chat), store directory, live offers, events, parking info.

Tenant — Footfall and revenue charts, AI business insights, maintenance requests, lease dashboard.

Admin — Mall KPIs, tenant management, AI strategic suggestions, complaints handling, announcements.

How we built it

React + Vite for the frontend, Tailwind CSS for styling, Recharts for data visualization, React Router v6 for navigation, and Google Gemini API (gemini-1.5-flash) for all AI features. Role-based auth using React Context and local Storage. Deployed on Vercel — no backend required.

Challenges we ran into

Designing three completely different dashboards inside one codebase without it becoming messy. Making Gemini give mall-specific answers instead of generic ones. Getting the app fully responsive on mobile. Deploying on Vercel with environment variables set correctly.

Accomplishments that we're proud of

Built a three-portal production-ready app in under 48 hours. Gemini integration that gives real, actionable insights. Clean UI that actually looks like a real product. Entire app runs serverless on Vercel at zero cost.

What we learned

How to call Gemini API directly from React without a backend. How to design role-based experiences with shared React Context. How to move fast with AI coding tools while keeping the code clean.

What's next for Zivra

Supabase auth, live IoT footfall data, WhatsApp alerts for tenants via Twilio, QR-based shopper check-in, and a pilot with a real mall operator in Pune.

Built With

• google-gemini
• react
• recharts
• tailwindcss
• vercel
• vite