This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass
What I Built
Gardenly — an AI-powered weekly garden planner that turns your garden into a simple, actionable outdoor routine.
Users enter their location, garden size, and plants, and Gardenly creates a personalized 7-day plan with tasks like soil preparation, watering checks, plant care, pest observation, and harvesting.
It combines garden context, seasonal information, and weather data to make the plan more relevant than a generic gardening chatbot. The app also includes plant scanning, task tracking, offline/local-first support, and a clean dashboard designed to make gardening easier and more consistent.
Demo
Code
Garden Week Planner
Local-first gardening MVP. The React client remains useful offline with a persisted garden and deterministic weekly plan; the FastAPI service adds Open-Meteo context, SQLite persistence, and optional Qwen structured planning through Hugging Face.
Run
Frontend: npm install then npm run dev.
Backend: cd backend, python -m venv .venv, activate it, then pip install -r requirements.txt and uvicorn app.main:app --reload.
Optional configuration: HF_TOKEN enables the Hugging Face Inference Providers integration with Qwen/Qwen3-4B-Instruct-2507:fastest. When it is absent or the service response is invalid, the app uses the rule-based planner.
How I Built It
Gardenly is built with React + Vite on the frontend and Python + FastAPI on the backend, with SQLite for local garden data.
The AI planning layer is designed around the open-weight Qwen3-4B-Instruct-2507 model through Hugging Face Inference Providers. The model receives the user's garden context—location, garden size, plants, and planning needs—and generates a structured 7-day outdoor plan.
I also added validation and a rule-based fallback, so the app remains functional when AI inference is unavailable. Weather context is integrated through Open-Meteo, and the app follows a local-first approach for storing garden data.
Why Does Open Innovation Matter?
Open innovation makes Gardenly more transparent, adaptable, and accessible. Using an open-weight model means the AI planning layer is not tied to a single closed AI provider.
It gives developers more control over how the model is used, how garden data is processed, and how the system can be improved or self-hosted in the future. It also makes it possible to build AI tools that can remain useful even when a particular provider or API is unavailable, thanks to Gardenly's fallback design.
For a gardening tool, this matters because the goal is to make personalized outdoor guidance accessible, flexible, and not dependent on one closed platform.
-Thank You!
Top comments (0)