I built TouchGrass Quest, a simple web app that encourages people to take a break from their screens and explore the outdoors.
Users can choose a location, select how much time they have, and pick an activity mood. The app generates a small outdoor quest to help them observe nature, move around, relax, or get creative.
I wanted to build something simple that could turn a few minutes away from the screen into a small real-world adventure.
This was my first Hacktoberfest project, and I focused on making something lightweight that people can use directly in their browsers.
Demo
Try the live app here:
https://huggingface.co/spaces/yashkhokhar/TouchGrass_Quest
Code
The source code is available in my Hugging Face Space:
https://huggingface.co/spaces/yashkhokhar/TouchGrass_Quest/blob/main/index.html
The project is built with HTML, CSS, and JavaScript.
How I Built It
I built the interface using HTML, CSS, and JavaScript and hosted it using Hugging Face Spaces with the Static SDK.
For the AI functionality, I used Transformers.js with the Xenova/flan-t5-small model. This allows compatible model inference to run in the browser without setting up a separate Python server or paid inference API.
The project taught me how to connect a browser interface to an open-weight language model and handle model loading and generated responses.
The model's output is still basic, and there is plenty of room to improve the quality of the generated quests.
Why Does Open Innovation Matter?
Open innovation gave me an opportunity to experiment with AI without needing to run a large model on my own computer or depend on a paid AI API.
Using an open-weight model and an open-source browser inference library made it possible for me to explore how AI works in a small web application.
As a beginner, I found it valuable to learn by building something practical, even with limited hardware and experience.
This project is an early step in my journey into open-source AI, and I hope to improve it further.
My Agent Session
I don't have an agent session to share for this project.
Prize Categories
No partner-specific category claimed.
Top comments (0)