Touch Grass

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

What I Built

Touch Grass is an AI-powered outdoor mission generator designed to turn screen time into real-world experiences.

We spend a significant amount of time looking at screens, even when we have a few minutes available to step outside. I wanted to build something that uses AI for a different purpose: helping people stop scrolling and start exploring the world around them.

Touch Grass turns a user's available time, energy level, and interests into a personalized outdoor activity. Whether someone wants to take a short walk, observe birds, try gardening, or get some exercise, the goal is to make getting outside easier to start.

Rather than generating another conversation that keeps users on their screens, Touch Grass aims to give them a practical mission they can take into the real world.

Demo

Live demo: [INSERT YOUR WORKING APP URL]

Demo video: [INSERT YOUR VIDEO URL, IF AVAILABLE]

The demo shows how a user can enter their preferences, generate an outdoor mission, and use the suggestion as a starting point for an offline activity.

Code

GitHub repository: [INSERT YOUR PUBLIC GITHUB REPOSITORY URL]

The repository includes the application source code, setup instructions, and implementation details.

How I Built It

I built Touch Grass as a web application powered by Qwen2.5-7B-Instruct, an open-weight language model released under the Apache 2.0 license.

The application uses hosted model inference to generate outdoor activity suggestions from user preferences. The frontend collects the inputs and displays the resulting mission, while the AI integration generates the personalized content.

The main design principle is to keep the experience simple: provide a few preferences, get a useful suggestion, and go outside.

The project focuses on making AI output actionable rather than encouraging users to spend more time interacting with a chatbot.

Current implementation: Hosted inference means the application depends on the configured API provider and an internet connection. It is not presented as a fully offline AI application.

Why Does Open Innovation Matter?

Open innovation matters because developers should be able to experiment with AI models, inspect their capabilities, and build applications around their own ideas without being limited to a single closed model ecosystem.

Using an open-weight model such as Qwen gives developers greater flexibility in model selection and experimentation. It also creates a path toward future customization, evaluation, and potentially self-hosted inference where hardware and resources permit.

For Touch Grass, the important lesson is that an AI model does not need to power a complicated chatbot or generate endless content. It can support a small, focused application that encourages people to interact with the physical world.

Hosted inference still introduces provider dependencies, so this implementation does not claim the independence or privacy characteristics of a fully self-hosted deployment.

Open innovation makes experimentation more accessible and gives developers more options as their projects evolve.

My Agent Session

Optional: [INSERT YOUR DEVRELAY AGENT SESSION LINK, IF AVAILABLE]

Prize Categories

[ADD ONLY THE PARTNER CATEGORIES THAT APPLY, OR REMOVE THIS SECTION IF NONE APPLY]