Google Find

Inspiration

Have you ever lost something that mattered deeply to you?

In Japan, I lose my wallet about once a year.
Each time, someone finds it and returns it.
But when I moved to the U.S. and lost my passport… it never came back.

Worse than losing it was realizing how broken the lost-and-found system is here.
No one knows where to turn something in.
Even institutions don’t have a clear process.
And searching? It’s exhausting.

That moment of frustration inspired me to build something better.

What it does

Google Find is an AI-powered platform to solve lost-and-found.
It lets users register lost or found items with smart tagging: category, color, size, features, location, and date.
Facility staff can search easily. Users can track matches in real time.

And we reward finders—with platform tokens.
If an item is unclaimed after a year, it goes to auction.
The tokens can be used to redeem items or bought with real currency.

How we built it

We built a web-based prototype using:

Gemini API for natural language tagging and image understanding
Firebase for authentication and storage
Google Maps API for item location mapping
OpenAI Embeddings to match lost and found items via image + text similarity

We designed a simple UI for both general users and staff at airports or large venues.

Challenges we ran into

Structuring unstructured descriptions (e.g. "It’s a black tote with cat stickers")
Building effective, ethical reward systems for finders
Creating a UX that works for both individuals and institutions

Accomplishments that we're proud of

Successfully integrated AI to predict possible matches between lost and found items
Designed a token-based incentive model to promote honest returns
Created a working demo usable on both desktop and mobile