Touchgrass.AI

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

What I Built

I built TouchGrass AI, an open-source AI companion designed to help people spend less time on their screens and more time outdoors. 🌱

The idea is simple: use AI for a few seconds, then put the phone away.

Users can choose an activity mode, available time, energy level, outdoor setting, and personal interests. TouchGrass AI then generates a short, practical outdoor mission that they can actually go outside and complete.

The app currently includes three modes:

• 🌿 Nature — discover and observe things around you
• 🚶 Explore — generate small outdoor exploration challenges
• 🌱 Garden — get simple gardening and plant-care activities

It is designed for anyone who wants a simple reason to step outside, whether that's a short walk, exploring a local park, observing nature, or spending some time in the garden.

The goal isn't to make people spend more time interacting with AI.

The goal is to make the AI interaction as short as possible.

Code

💻 GitHub: [https://github.com/Varun-666/touchgrass.ai]

The repository contains the complete application, including the frontend, FastAPI backend, AI integration, prompts, fallback system, PWA support, tests, and documentation.

How I Built It

TouchGrass AI is built around open-weight AI and local inference.

The AI component uses Qwen2.5-0.5B-Instruct, accessed through the Hugging Face Transformers ecosystem.

The basic flow is:

User
  ↓
TouchGrass Web UI
  ↓
FastAPI Backend
  ↓
Qwen2.5-0.5B-Instruct
  ↓
Outdoor Mission
  ↓
Put the Phone Away
  ↓
Go Outside 🌳

The user provides a few preferences such as:

• Activity type
• Available time
• Energy level
• Outdoor setting
• Interests or goals

The local AI uses these inputs to generate a structured outdoor mission containing a title, description, steps, duration, things to bring, and safety information.

The project uses:

• Python
• FastAPI
• Hugging Face Transformers
• Qwen2.5-0.5B-Instruct
• HTML/CSS/JavaScript
• Progressive Web App support

There is also a fallback system so the application can still provide useful outdoor activities if the AI model is unavailable.

No paid AI API or API key is required.

Why Does Open Innovation Matter?

Open innovation is important to TouchGrass AI because the project is intended to work with people in the real world, not just inside a cloud-based chat window.

🔒 Privacy

Outdoor experiences can involve personal information such as photos, gardens, homes, and surroundings.

With local inference, the project can process AI requests without requiring everything to be sent to a third-party AI provider.

📡 Offline Potential

The places where people want to use TouchGrass AI may not always have a reliable internet connection.

Using an open-weight model gives the project a path toward running the AI locally and eventually supporting completely offline experiences.

🔧 Freedom to Experiment

The AI isn't locked to a proprietary API.

Developers can experiment with:

• Different open-weight models
• Smaller models
• Quantized models
• Fine-tuned models
• Different prompts
• Different local hardware

This makes the project much easier to modify and extend.

💰 Accessibility

There are no per-request API costs.

Once the model is available locally, users can experiment with the application without needing a paid AI API subscription.

For me, that's one of the biggest advantages of open AI: the technology becomes something developers can actually take apart, modify, and build on.

My Agent Session

[Add your DevRelay agent session here if available]

Prize Categories

• Touch Grass
• Open-Source AI

---

Less screen. More outside. 🌱