This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass.
🌿 What I Built
Most productivity apps ask us to spend even more time looking at a screen.
I built TouchGrass AI, an outdoor mission planner designed to do the opposite.
You tell it:
- ⏱️ How much time you have
- 🚶 What kind of activity you want
- ⚡ Your current energy level
TouchGrass AI creates a simple outdoor mission with a step-by-step plan and a phone-down challenge.
The goal is simple: use the screen for a few seconds, get your mission, put the phone away, and go outside.
🌎 How It Gets People Outside
Instead of giving users another feed, tracker, or endless list of activities, TouchGrass AI gives them one practical mission.
For example:
Take a 30-minute walk, explore your surroundings, keep your phone in your pocket, and notice three things you normally miss.
The generated mission is designed around the user's available time, activity preference, and energy level.
The screen is meant to be the shortest part of the experience.
🚀 Demo
🌐 Live Demo:
https://harshitha-hg.github.io/Touchgrass-ai/
Try selecting different time, activity, and energy combinations and generate an outdoor mission.
💻 Code
🔗 GitHub Repository:
https://github.com/Harshitha-HG/touchgrass-ai
The project is completely public so anyone can inspect, fork, modify, and improve it.
🛠️ How I Built It
TouchGrass AI is built with:
- HTML
- CSS
- JavaScript
- WebLLM
- SmolLM2 open-weight model
- GitHub Pages
The AI model runs through WebLLM directly in the browser rather than sending every request to a closed AI API.
This makes the open model an important part of how the project works, rather than simply using AI as an add-on.
I also added a lightweight fallback so the outdoor-planning experience remains usable when local AI inference cannot return a response on a particular device.
🔓 Why Open Innovation Matters
For this project, open AI makes experimentation much more accessible.
With an open-weight model and browser-based inference, developers can inspect the project, experiment with the model, change the prompts, and build their own version without depending entirely on a closed API.
It also creates an interesting possibility for privacy-focused applications: the long-term goal is to make more of the experience run locally on the user's device rather than constantly sending personal interactions to a remote service.
Most importantly, open innovation means this small idea doesn't have to remain my idea. Someone can fork TouchGrass AI, replace the model, improve the missions, add local languages, add weather information, or turn it into a completely different outdoor experience.
🌱 What I Learned
The biggest lesson was that an AI project doesn't have to keep people staring at AI.
AI can be the small bridge between "I don't know what to do" and "I'm outside doing it."
That is the idea behind TouchGrass AI.
Use the screen to get the mission. Then touch grass. 🌿
Top comments (0)