LostLink

Inspiration

LostLink is a personal-loss-prevention and lost-&-found community platform built to solve two closely related everyday problems:

Forgetting items when you leave a place — We all walk away from cafés, gyms, offices or friends' houses and only realise 20 minutes later that we left a bag, keys, wallet or laptop behind. LostLink prevents this at the source by turning your phone into an intelligent companion that knows when you are leaving a location and what you should have with you.

Reconnecting people with lost belongings — When an item is already lost, the traditional process (posters, social-media posts, police reports) is fragmented and inefficient. LostLink creates a structured, privacy-respecting marketplace where the person who lost an item and the person who found it can be automatically matched and communicate safely.

What it does

Real-Time Movement Tracking & Smart Alerts LostLink continuously monitors your physical activity state using the phone's accelerometer and gyroscope (via the browser DeviceMotionEvent API). It detects steps like a dedicated step-counter, classifies your movement as stationary, walking, running or in a vehicle, and derives an estimated speed from step cadence. A built-in stationary alert arms itself after you have been still for a user-configurable threshold (default 2 minutes). The moment you transition from stationary to walking, the app vibrates, fires a push notification and shows an in-app toast asking “Did you forget something?” — prompting you to check your belongings before you get too far away.

Smart Location-Based Checklists Users create checklists tied to specific geographic locations and time windows (e.g. “Gym bag — weekdays 6:00–9:00 at my gym coordinates”). While a checklist is active, LostLink compares your live GPS position against the checklist location. If you move more than 50 metres away from that location while the checklist is active, the app immediately vibrates and opens a departure reminder modal telling you exactly which checklist you are leaving behind. Once you return within 30 metres, the alert resets so it can fire again on the next departure.

Lost & Found Matching When a user reports a lost item (with name, category, description, photos, estimated loss location and time) or a found item, LostLink runs a custom matching algorithm through a Supabase Edge Function. The algorithm computes a composite similarity score based on four weighted factors:

Text similarity (40%) — keyword overlap between descriptions Location proximity (30%) — Haversine distance (closer is better, up to a 5 km radius) Time proximity (20%) — how close the found time is to the estimated loss time (up to 48 hours) Category match (10%) — overlap between the lost-item category and the found-item object type Matches scoring 30% or above are surfaced to both parties on the Matches screen. Each match creates a private chat room so the owner and finder can communicate securely without exposing personal contact details.

Secure In-App Chat with Ratings Every match gets its own ephemeral chat room created on-demand via a Supabase Edge Function that enforces participant identity. Real-time messaging is powered by Supabase's Realtime Postgres change subscriptions. After an item is successfully recovered, the searcher can rate the finder/helper on a 1–5 star scale, building a community reputation score and total recoveries count that is visible on volunteer helpers' profiles.

Route History & Analytics All movement data (GPS coordinates, speed, movement type) is persisted to the database. Users can open the Route History page to visualise their past movements on an interactive Leaflet map, colour-coded by movement type, with overlaid stop markers and summary statistics (total distance travelled, number of stops, average speed).

How we built it

We built it by smartly iterating with MeDo in our conversation. We described it what to do and it just did it. If an error occured, we gave it the content of the error message and a description of what happened and it fixed it. To be more rapid, when we had many little improvements to do, instead to send a prompt for each of them, we made a structured prompt with all the modifications we want, gaining in time.

Challenges we ran into

To explain properly to the AI what we want and make it understand it clearly.

Accomplishments that we're proud of

The features we built with MeDo and whose we are really proud are:

• alert and reminder when moving again after stopping during a certain amount of time
• smart checklist with reminder based on days, hour an location
• AI automatic matching between found and lost item
• in-app chat between finders and searchers ## What we learned

What's next for LostLink

Built With

• medo