GoTrail: AI Nature Discovery App for Hikers

#Health

#Outdoors

Event

LA Hacks

Team

Austin Lieu, Harvey Tan, Ryan Longacre

GoTrail is a native iOS app built at LA Hacks 2026 that turns hiking into a Pokédex-style nature discovery game. Built by Austin Lieu, Harvey Tan and Ryan Longacre, GoTrail won the Figma Make Challenge by combining on-device plant identification with gamified trail exploration to make hiking more engaging and accessible for beginners.

What it solves

Doctors prescribe hiking and time in nature for mental and physical health more than almost any other lifestyle intervention. But adherence is famously low. Most people find hiking boring and beginners feel intimidated in unfamiliar environments surrounded by plants and species they cannot identify. GoTrail fixes both problems by making every hike a discovery experience and giving doctors a tool they can actually prescribe.

What it does

Users start a hike, point their camera at any plant, and GoTrail identifies it in under a millisecond with zero internet connection. Every identification is saved to a personal collection with its photo, location and timestamp, building a living map of every trail the user has explored. Hiking statistics, species collections and an interactive trail map combine to create a portfolio that keeps users coming back outside.

How it works

GoTrail uses ZETIC Melange to deploy a plant classification model trained on the PlantNet-300K dataset of 300,000 botanical images directly on the device. Inference runs locally at millisecond-level latency with no network calls, making the app fully functional deep on trails with no signal. The app is built in Swift and SwiftUI with AVFoundation for camera control, MapKit and CoreLocation for trail mapping, and SQLite for offline-first local storage. Supabase syncs the collection to the cloud only when wifi is available.