Touch Grass, Meet Birds: A Local AI Balcony Sanctuary Planner to Get You Off the Screen

Ayush Jaiswal — Sun, 11 Oct 2026 17:18:25 +0000

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

What I Built

Modern urban life often traps us in front of screens inside concrete apartments, disconnected from the living world just outside our windows. Balcony Birds is a privacy-first, lightweight interactive web guide designed to help urban apartment dwellers transform balconies of any size into micro-sanctuaries for local wildlife.

Instead of keeping users hooked to an endless digital feed, Balcony Birds has one explicit goal: get you off your computer and out onto your balcony.

You tell it where you live (across Indian regions), your balcony's sunlight, pot space, and floor level. It computes the season, determines which local bird species can realistically visit your floor, recommends container plants that provide nectar/seeds/cover, provides carpenter-grade blueprints for DIY nest boxes, and compiles an actionable weekly outdoor to-do list.

Once your plan is generated, hit Print the sheet, close your laptop, and step outside into the fresh air with your physical checklist.

Who It's For:

Urban apartment residents and balcony gardeners wanting to invite nature back into high-density spaces.
Beginners who want to start birdwatching without expensive gear.
Families and hobbyists looking for a weekend DIY nature project.

Demo

Live Application: [https://github.com/AYUSHJAISWAL12123/Birds]

- Zero Setup Required: Works immediately in any modern web browser—no Node.js, bundlers, or servers needed.

Code

Single-File Architecture: Built entirely with semantic HTML5, modern CSS with SVG animations, and vanilla JavaScript.
No External Dependencies: Zero npm packages, zero external trackers, and zero build pipelines.
MIT Licensed: Completely free to study, fork, modify, and expand.

How I Built It

1. Grounded Deterministic Engine + Local Open-Weight AI

Rather than letting an LLM hallucinate biological seasons, nesting dimensions, or plant care:

A deterministic JavaScript engine first evaluates the user's region, sunlight, floor height, and current month against a curated regional dataset of urban Indian birds (House Sparrows, Magpie-Robins, Purple Sunbirds, Tailorbirds, Mynas, Bulbuls, and Parakeets).
The engine generates structured nesting dimensions (entrance hole diameter in centimeters, depth, floor area) and container sowing seasons.
For personalized insights, Balcony Birds integrates with Ollama running locally on consumer hardware (http://localhost:11434/api/generate) using open-weight models such as qwen3:4b (or Llama 3 / Mistral).
The plan is passed with a grounded system prompt that instructs the model to act as a warm neighborhood gardening guide, explaining why each plant supports each visiting bird, while strictly forbidding it from making up unverified facts.

2. Immersive Visual Design

Animated Sky Backdrop: Pure CSS animations and custom SVG paths generate drifting clouds and a layered flock of flying birds moving at varying altitudes and speeds across a daytime sky gradient.
High-Contrast Minimalist Cards: Sleek, borderless black cards ensure comfortable readability while keeping the focus on nature.
Print Stylesheet: Custom @media print rules strip out forms, controls, and animations, leaving only a clean, ink-friendly sheet formatted for clipboard use outdoors.

Why Does Open Innovation Matter?

Connecting with nature shouldn't require a monthly subscription, an API credit card, or sending personal habits to proprietary cloud silos.

Accessibility & Equity: Open-weight models (running via Ollama) democratize AI so that students, teachers, and conservationists can run intelligent offline software anywhere in the world, completely free.
Total Privacy: When your gardening and lifestyle data runs against a local open model on your own device, nothing ever leaves your computer.

3. Local Adaptation & Community Ownership: Nature is hyper-local. A closed proprietary API will never prioritize the specific nesting needs of urban Indian birds or local potting soil constraints. With open-source code and open weights, local bird societies, schools, and botanists can freely fork the repository, localize the species database for their own bio-region, and tune the model on their own terms.

Prize Categories

Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass
Local Inference / Open-Weight AI

DEV Community: Ayush Jaiswal