Verdant AI — Intelligence Rooted in Nature

Harshit Bajpai — Fri, 09 Oct 2026 11:13:45 +0000

🌿 Verdant AI — Intelligence Rooted in Nature

LIVE WEBSITE - https://rage-arch.github.io/verdant-AI/

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

What I Built

Verdant AI is an AI-powered nature companion designed to help people spend less time scrolling and more time exploring the world around them.

In a world where we spend hours looking at screens, Verdant AI turns technology into a bridge to the outdoors rather than a distraction from it.

🌱 What it offers:

AI Nature Guide: Ask questions about plants, ecosystems, leaves, and the natural world using a locally running open-weight AI model.
Outdoor Field Missions: Complete seven screen-free challenges that encourage you to observe leaves, listen to nature, notice clouds, explore colors, and discover local habitats.
Digital Field Journal: Record discoveries and observations, track completed missions, and export your journal as a JSON file.
Privacy-Focused Experience: Journal entries are stored locally in your browser, while the AI guide can run on your own computer without a hosted AI API.
Offline-First Design: After the necessary setup and model download, the local AI can work without an internet connection, provided the app server and model are running.

Verdant AI is designed for students, nature enthusiasts, curious beginners, and anyone who wants a simple reason to step outside and reconnect with their surroundings.

The goal is simple: use AI to encourage real-world experiences, not replace them.

Demo

🌐 Live Website: https://rage-arch.github.io/verdant-AI/

💻 GitHub Repository: https://github.com/RagE-arch/verdant-AI

The hosted website showcases the interface. The full local AI chat functionality requires running the project with Node.js and Ollama, as explained in the repository's README.

Code

Explore the complete source code in the public repository:

Verdant AI on GitHub

The project is built with:

HTML for the application structure
CSS for responsive layouts and visual design
JavaScript for missions, navigation, journaling, and AI interactions
Node.js for the local server and AI request handling
Ollama for running the AI model locally

The repository includes the application files, setup instructions, service worker, web app manifest, and MIT license for the app code.

How I Built It

The core of Verdant AI is Qwen2.5-1.5B-Instruct, an open-weight language model that runs locally through Ollama.

Instead of relying on a hosted AI API, I designed the application around a local architecture:

Frontend: HTML, CSS, and JavaScript create the dashboard, nature guide, outdoor missions, and field journal.
Local backend: A Node.js server serves the application and handles chat requests.
AI inference: The server sends questions to the local Ollama API, which runs the configured Qwen model.
Local data storage: The field journal and mission progress are saved in browser storage.
Offline support: A service worker caches the application shell after the first visit, while Ollama manages the downloaded model separately.

The app also uses a nature-focused system prompt that encourages concise educational answers, acknowledges uncertainty, and discourages unsafe interactions with unfamiliar plants and wildlife.

One important distinction: the first setup requires downloading the software and model. Afterward, local AI inference can work offline while the necessary local services are running.

Why Does Open Innovation Matter?

Open innovation matters because an application about reconnecting with nature should not require constant connectivity or send every question to a remote AI provider.

Using an open-weight model and local inference makes several things possible:

More control: Developers can inspect the architecture, change the model, and customize the AI's behavior.
Privacy by design: Questions can be processed locally, and journal entries remain in browser storage rather than being uploaded to a hosted journal service.
Reduced dependency: The local AI guide does not require a paid hosted AI API key.
Offline potential: Once the model and software are installed, the AI can continue working without an internet connection.
Community collaboration: Open-source code allows other developers to study, adapt, improve, and extend the project.

A closed API could provide similar conversational features, but it would introduce a dependency on a remote provider and its availability, pricing, and data-handling policies.

For Verdant AI, openness is not just a development choice. It supports the project's goal of making nature education more accessible, adaptable, and privacy-conscious.

My Agent Session

I have not included a DevRelay agent session in this submission.

The source code and README are available in the GitHub repository for reviewing the implementation and setup process.

Prize Categories

Overall Challenge: Open-Source AI — Touch Grass

I am submitting Verdant AI for the overall challenge.

I am not claiming a partner-specific prize category at this time, as the current implementation uses Qwen2.5 through Ollama rather than one of the listed partner technologies.

🌎 Final Thought

Technology doesn't always have to keep us indoors.

Sometimes, the best thing an AI can do is answer a question, suggest something worth noticing, and encourage us to put the phone away.

Verdant AI — Intelligence rooted in nature. 🌿

Hacktoberfest #OpenSourceAI #AI #Qwen #Ollama #WebDev #BuildInPublic #TouchGrass

DEV Community: Harshit Bajpai