This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass
What I Built
TouchGrass AI turns a nearby starting point, the time you have, and your mood into a short outdoor micro-adventure. It gives you three simple steps, then encourages you to put your phone away and go do them.
It’s for anyone who wants a quick reason to take a walk, explore a nearby park, or notice more of the nature around them.
Demo
Try TouchGrass AI
Code
View the source code on GitHub
How I Built It
I built TouchGrass AI with HTML, CSS, and JavaScript. It uses the open-source WebLLM runtime to run the open-weight Qwen2.5 1.5B Instruct model in the browser using WebGPU.
The user’s place, available time, and selected mood become a prompt for the model. It generates a short plan with a title, an introduction, and three steps. The first model download requires internet; after setup, inference runs locally in the browser.
Why Does Open Innovation Matter?
TouchGrass AI is meant to help people spend less time on screens, so I wanted the AI generation to work without sending their outing details to a hosted AI service. Local inference keeps the prompt on the user’s device and avoids needing an AI API key or paying per request.
Using open tools also makes the project easier to adapt: I can change the prompt, try a different supported model, and shape the guide around the experience I want to build.
My Agent Session
I don’t have a DevRelay session link to share.
Prize Categories
Open-Source AI Challenge — Touch Grass.
Top comments (0)