Plan quickly. Go outside. Touch grass. 🌱
Students often decide they want to go outside, but spend more time deciding what to do than actually doing it.
For the Hacktoberfest 2026 Open-Source AI Challenge: Week 1, themed Touch Grass, I built Ranchi EcoRoute — a small AI-powered outdoor activity planner that turns a few seconds of planning into a practical activity plan.
The idea is simple:
Use the screen briefly to make a plan, then put the phone away and go outside.
🤔 What I Built
Ranchi EcoRoute asks the user for:
- Outdoor activity
- Available time
- Energy level
- Optional custom activity
It then uses Google's Gemma open-weight model to generate:
- A suggested outdoor activity
- A suitable type of place
- A time-based activity plan
- Things to carry
- Basic safety tips
- A screen-break goal
The project is intentionally simple because the screen should be the shortest part of the experience.
🧠 How It Works
User Input
(Activity + Time + Energy)
|
v
Google Gemma
|
v
Outdoor Activity Plan
|
v
Suggested Place Type
+ Time-based Plan
+ What to Carry
+ Safety Tips
+ Screen Break Goal
|
v
Gradio UI
Gemma handles the natural-language generation of the outdoor plan.
Python validation and prompt constraints are used to keep the generated plan aligned with the user's available time and energy.
The project does not claim to provide live weather, traffic, opening hours, or real-time route information.
🌱 The Touch Grass Idea
The core idea behind the project is:
Less screen time
↓
Quick plan
↓
Go outside
↓
More real-world activity
Instead of building another tool that keeps users inside an app, I wanted the app to make itself unnecessary as quickly as possible.
🖥️ Demo
The prototype runs in Google Colab and launches a Gradio web interface.
Example
Outdoor Activity:
Morning Walk
Time Available:
45 minutes
Energy Level:
Low
Gemma generated a structured plan with a short walking schedule, suggested place type, things to carry, safety tips, and a screen-break goal.
Live Demo
⚠️ The Gradio share link is temporary. The source code and screenshots are available in the GitHub repository.
🧪 Testing
I tested the prototype with multiple activity configurations.
Morning Walk
Input:
45 minutes
Low energy
The model generated a short nature-oriented walking plan suitable for the available time and energy level.
Nature Photography
Input:
60 minutes
Medium energy
The model generated a photography-focused outdoor plan with a time breakdown, things to carry, safety tips, and a screen-break goal.
These tests helped verify that the generated plan changes according to the user's activity, available time, and energy level.
🌱 Why Open Innovation & Gemma Matter
Ranchi EcoRoute uses Google's Gemma, an open-weight model, as the core AI component.
I wanted the model to handle the actual natural-language planning rather than simply placing an AI chatbot inside a conventional interface.
An open-weight approach also leaves room for future experimentation such as:
- Swapping models
- Adapting the model for local use cases
- Exploring local or offline inference
- Running the model in more constrained environments
The interesting part is that the AI is used for planning, while the real goal happens away from the screen.
📸 Screenshots
Morning Walk
Nature Photography
📚 What I Learned
The biggest lesson from this project was that a useful AI application does not need to be complicated.
A small set of inputs, a focused prompt, and a simple interface can turn a general-purpose model into a useful tool for a specific real-world goal.
I also learned that an AI product can be more meaningful when its purpose is to reduce screen time rather than increase it.
🔮 Future Improvements
- Real-time weather integration
- Verified local place information
- Map-based route suggestions
- Offline/local Gemma inference
- Personalized activity history
- Better accessibility support
- Group activity planning
- More detailed outdoor difficulty levels
📁 Source Code
GitHub:
https://github.com/Avinash-sdbegin/ranchi-ecoroute
The repository contains the Colab notebook, README, and screenshots.
⚠️ Safety & Limitations
This is an educational and hackathon prototype.
It does not provide live weather information, real-time traffic information, live route tracking, verified opening hours, or medical advice.
Always verify weather, location access, local conditions, and personal safety before heading outdoors.
Built for the Hacktoberfest Open-Source AI Challenge: Week 1 — Touch Grass 🌿
Top comments (0)