I Built TouchGrass AI: An Open-Source AI That Encourages You to Go Outside 🌿

I built TouchGrass AI, an open-source AI web application that encourages users to spend more time outdoors rather than staring at screens. The goal of this project is to use AI to motivate people to step away from technology and engage with the natural world. TouchGrass AI allows users to select an outdoor activity, the duration they have, and their current mood.

Based on these inputs, the AI generates a short outdoor challenge, such as walking through a park or observing nature. Users can then mark the challenge as completed, request a new challenge, or start a brand new one. The application utilizes the open-weight Qwen2.5-0.5B-Instruct model via Transformers.js, enabling it to run directly in the browser without relying on expensive AI APIs.

The project was developed using Next.js, React, TypeScript, Tailwind CSS, Transformers.js, and the Qwen2.5-0.5B-Instruct model. The AI is designed to generate concise ideas rather than lengthy responses to keep the user experience focused on outdoor activities. The main inspiration behind TouchGrass AI is to explore whether AI can inspire individuals to disconnect from their devices and enjoy the outdoors more.

By leveraging open-source AI technology, the application demonstrates how it can be seamlessly integrated into web applications to promote real-world engagement.