My new project

#devchallenge #hf26challenge

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

🌿 TrailSpark AI — One Minute of AI, Then Go Outside

What I Built

TrailSpark AI is a small outdoor activity planner designed to use AI for the opposite of what we usually expect from AI: helping people leave the screen.

You tell TrailSpark:

📍 Where you are
⏱️ How much time you have
⚡ Your energy level
👥 Who you're with
🌳 What sounds fun
🚫 Any constraints

It then turns those preferences into a short outdoor mission with a simple goal, something to notice, a playful challenge, and a stopping point.

The idea is simple:

Use AI for one minute → close the screen → go outside.

The project is aimed at students, remote workers, and anyone who wants a small push to spend more time outdoors without turning it into another complicated planning session.

Demo

🌐 Live Demo:

https://trailspark-lprbvtdsa-adityaraj3532.vercel.app/

Try entering your preferences and clicking Generate my mission.

The first generation can take a little longer because the open-weight model is downloaded and cached by the browser.

Code

💻 GitHub Repository:

https://github.com/adityaraj3532/trailspark-ai

The project is a static web application built with HTML, CSS, and JavaScript.

How I Built It

The most important part of TrailSpark is that the AI runs locally in the browser.

I used:

Xenova/flan-t5-small — an open-weight model
Hugging Face Transformers.js — to run the model in the browser
HTML, CSS, and JavaScript for the application

The basic flow is:


text
User preferences
       ↓
Prompt template
       ↓
Xenova/flan-t5-small
       ↓
Local browser inference
       ↓
Outdoor mission
       ↓
📵 Close the screen
       ↓
🌿 Go outside

DEV Community