🌿 TrailMate AI — Your AI Companion for Getting Outdoors

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

What I Built

🌿 TrailMate AI — Your AI Companion for Getting Outdoors

As students, we spend so much time on our phones and laptops that we sometimes forget to step outside and experience the world around us.

That's why I built TrailMate AI, a simple outdoor adventure companion that encourages students to take a break from screens and reconnect with nature.

Users choose their available time, preferred difficulty, and type of outdoor environment. TrailMate AI then generates small outdoor missions — exploring a nearby park, observing birds, taking a mindful walk, or discovering something new in nature.

The app is designed to make outdoor activities easy to start, even when you only have five minutes.

The goal is simple: less scrolling, more exploring, more real-world experiences.

Demo

🌐 Live Demo: https://trailmate-ai-ypcf.onrender.com

First visit needs internet to download the local AI model (~711 MB). After that, everything runs offline.

Code

💻 GitHub Repository: https://github.com/ath-il/trailmate-ai.git

How I Built It

TrailMate AI is built with React, TypeScript, and Vite, with a lightweight, responsive interface.

For local AI it uses WebLLM to run Google's open-weight Gemma 3 1B model directly in the browser through WebGPU — no paid cloud AI API. Mission generation happens on the user's own device.

The app is offline-first: a service worker (Vite PWA / Workbox) caches the app shell, journal entries live in IndexedDB, and bundled hand-authored missions keep it useful when AI inference isn't available. The first run downloads the model once; after that, the app works with no connection.

The static front end is deployed on Render via a committed render.yaml Blueprint.

Why Does Open Innovation Matter?

I believe AI should empower people rather than make them more dependent on screens or centralized services.

Open-weight models give developers the freedom to experiment, swap models, and explore local inference without relying entirely on proprietary AI APIs.

For TrailMate AI this offers three benefits:

• Privacy — mission preferences are processed locally instead of being sent to a cloud provider.
• Accessibility — after the initial setup, users generate missions with no ongoing API charges.
• Freedom to experiment — developers can try different compatible models without lock-in.

There's also something meaningful about using AI to encourage people to put their phones down and experience the world around them.

The best technology doesn't always need to keep us online. Sometimes it should help us disconnect.

Prize Categories

• Best Use of Gemma — Google's open-weight Gemma 3 1B is the default model, running fully on-device via WebLLM/WebGPU.
• Best Use of Render — the static app is deployed on Render using the included render.yaml Blueprint.