Lost2Found

Inspiration

Lost2Found was inspired by a real experience. I once lost my passport at a visitor center bookstore while traveling. By the time I realized it was missing, I had already visited multiple places. There was no single system where I could report it once and check across all the locations I had been to. I had to physically retrace my steps, go back to each location, and ask every lost-and-found desk separately.

That stressful experience made me realize that the problem is not only losing something important. The bigger problem is that lost-and-found systems are still disconnected, manual, and location-based. If you lose a passport, wallet, ID, laptop, or keys, you are forced to become the search engine yourself.

Lost2Found was created to make that process faster, smarter, and less stressful.

What it does

Lost2Found is an AI-powered lost-and-found network that helps people recover lost items without physically retracing every place they visited.

Users can report a lost item once, add details about the item, and select the locations they visited. Venues, bookstores, visitor centers, campuses, event organizers, or front-desk staff can report found items from their side. Lost2Found then uses AI to compare lost and found reports using item descriptions, photos, time, location, and private identifying details.

When the system finds a likely match, it shows a confidence score and starts a safe claim process. The owner must verify hidden details before the item can be marked ready for pickup.

The goal is simple:

Report once. Search every place you visited.

How we built it

We built Lost2Found as a full-stack web application designed for both users and venue admins.

The frontend was built with a clean, simple interface so someone who is stressed after losing an important item can quickly submit a report. Users can create a lost report, upload an image if they have one, describe the item, select the places they visited, and add private proof details that only the real owner would know.

On the admin side, venues can create found item reports, manage active items, review AI match suggestions, and approve or reject claims.

For the backend, we designed the data model around real lost-and-found workflows: users, locations, lost reports, found reports, AI-generated matches, claim verification, and resolution status. The database acts as the source of truth for connecting reports across different locations.

The AI layer helps extract useful details from item descriptions and images, compares reports semantically, and generates match confidence based on item type, color, location, time window, and private proof signals.

Challenges we ran into

One of the biggest challenges was making the idea useful without making it unsafe. Lost-and-found has a trust problem: if a found item is shown with too many details, someone could falsely claim it. But if the system hides too much information, the real owner may not recognize it.

To solve this, we designed a private verification flow. Public reports only show limited information, while sensitive details are hidden and used for owner verification.

Another challenge was narrowing the scope. Lost-and-found can apply to airports, campuses, public transit, hotels, museums, events, and cities. Instead of trying to solve everything at once, we focused on a realistic first use case: venues and events where people can report lost or found items through a shared system.

We also had to think carefully about how AI should be used. We did not want AI to make final ownership decisions. Instead, AI helps suggest likely matches, while the final claim still depends on human verification and admin approval.

Accomplishments that we're proud of

We are proud that Lost2Found solves a problem that feels small until it happens to you. Losing something like a passport or wallet can create panic, wasted time, and real consequences. Our project turns that stressful process into a guided recovery flow.

We are also proud of building a product that is practical and deployable. Lost2Found does not require an entire city to adopt it on day one. It can start with one hackathon, one campus, one visitor center, or one event. A venue can share a QR code, users can submit reports, and admins can manage found items from a simple dashboard.

Most importantly, we are proud of the trust-first design. Lost2Found is not just about matching items quickly. It is about helping the right item return to the right person safely.

What we learned

We learned that a good solution is not always about creating a completely new behavior. Sometimes the best opportunity is improving a broken real-world workflow that already exists.

Lost-and-found systems are everywhere, but most of them are still manual, fragmented, and hard to search. By connecting locations and adding AI-assisted matching, we can make the process much more useful without replacing the human judgment that is still needed for trust and safety.

We also learned that AI is most valuable when it supports decision-making instead of pretending to be perfect. In Lost2Found, AI helps identify possible matches, but privacy, verification, and admin control remain central to the experience.

What's next for Lost2Found

Next, we want to make Lost2Found easier to deploy in real locations. Our goal is to support venue workspaces where campuses, conferences, visitor centers, bookstores, libraries, gyms, and coworking spaces can manage their own lost-and-found system.

Future features include route-based search, where users can enter all the places they visited and search across them at once; QR codes for quick venue reporting; automatic duplicate detection; multilingual support for travelers; and stronger privacy tools for sensitive items like passports, IDs, wallets, and medical bags.

Long term, Lost2Found could become a connected recovery network across places people visit every day.

Our vision is simple:

No one should have to physically retrace every step just to find something that was already turned in somewhere.

Built With

• amazon-web-services
• aurora
• node.js
• openai
• typescript
• vercel