Wayfin — Intelligent AI Indoor Navigation

Scan a QR and ask Wayfin anything. Conversational AI turns intent into reliable indoor routes across malls, airports, hospitals and campuses—with visual 3D guidance in an instant, no-download PWA.

Comment

Inspiration

Venues already have kiosks, display screens, and website maps—but they are static, easy to miss, and make visitors interpret the space themselves. Wayfin puts the experience in a visitor’s hand: scan a QR code, ask naturally, and get a personalized visual route instantly.

What it does

Wayfin is an AI indoor navigation PWA for malls today and airports, hospitals, and campuses tomorrow. Users can ask, “Take me from P1 to UNIQLO,” “Find a birthday dress,” “Where did I park?” or “Find my friend.” It returns a floor-aware route with 3D visual guidance, landmarks, parking, and store discovery—without an app download.

How we built it

I built Wayfin with React, Vite, a mobile-first PWA, CSS/SVG 3D maps, and a deterministic indoor routing graph. OpenAI’s Responses API converts natural language into strict structured intent; the routing engine then calculates the route. AI understands the user, while deterministic logic ensures directions are reliable.

Challenges we ran into

Indoor spaces do not have dependable GPS. We had to model floors, lifts, parking zones, landmarks, and store coordinates while keeping the experience clear on a phone. We also addressed the gap between an ambitious AR vision and what mobile browsers can reliably localize today.

Accomplishments that we're proud of

  • Removed app-download friction with QR-to-PWA access.
  • Turned passive venue maps into a conversational AI experience.
  • Combined visual 3D navigation with natural-language guidance.
  • Kept AI constrained to approved venue data instead of letting it invent routes.
  • Built one reusable architecture for many indoor-space categories.

What we learned

The best AI navigation experience is hybrid: AI should understand ambiguous human requests, but a deterministic system should calculate the route. We also learned that visual context—nearby stores, floors, landmarks, and “You are here”—creates more confidence than text directions alone.

What's next for Wayfin — Intelligent AI Indoor Navigation

Next, we will build a venue-management dashboard where malls, airports, hospitals, and campuses upload maps, coordinates, and live information. QR and visual anchors can re-localize users at entrances, lifts, and pillars. Longer term, native ARCore/ARKit support can enable truly anchored camera guidance.

Built With

Share this project:

Updates