What I Built
PlantGuard AI is an offline-first plant and leaf health journaling web application powered by Google Gemini AI. It is designed to bridge the gap between technology and nature by encouraging people to step away from their screens, step outside, and interact with the natural world.
Whether you're tending to a backyard garden, caring for house plants, or exploring local nature spots, PlantGuard AI helps you capture leaf photos, get honest AI-assisted diagnoses, track plant care habits with streaks, and log garden milestones.
To ensure it works seamless in real outdoor conditions (where mobile signal or Wi-Fi is often weak or non-existent), the app features an offline-first PWA architecture. Users can log observations, photos, and location points offline in IndexedDB and analyze them with Gemini whenever they reconnect.
Demo
Live Demo / Deployed Link: https://meetmittan.github.io/PlantGuard/
GitHub Repository: https://github.com/MeetMittan/PlantGuard.git
Code
https://github.com/MeetMittan/PlantGuard
How I Built It
PlantGuard AI strikes a balance between lightweight client-side performance and secure AI multimodal analysis:
AI Engine & Architecture: Powered by Google Gemini API (gemini-3.8-flash / gemini-2.0-flash) routed through a lightweight Node.js Express backend. This architecture keeps the GEMINI_API_KEY strictly on the server and avoids client-side device overheating or RAM saturation (0 MB download on user devices compared to heavy local vision models).
Frontend Stack: Built with Vite, React/TypeScript, and Tailwind CSS as a fully installable PWA (Progressive Web App).
Offline Storage: Utilizes IndexedDB to store observations, plant profiles, daily checklists, care streaks, and saved map locations locally on the device.
Mapping & Geolocation: Integrated Leaflet.js and OpenStreetMap to log outdoor nature exploration points without sending GPS data to third-party AI services.
Why Does Open Innovation Matter?
Open innovation and open-weight/transparent toolchains matter because plant health and user privacy should never be trapped inside closed black boxes.
By building PlantGuard AI on open web standards, modular architectures, and transparent AI prompt pipelines, we prove that powerful AI can be delivered responsibly. Open innovation allowed us to construct a solution that prioritizes privacy (no GPS or personal data sent to AI), battery conservation, and true offline accessibility—something proprietary walled-garden apps rarely offer.
My Agent Session
View Agent Session
Prize Categories
AI & Open Source Apps (Google Gemini API integration)
Touch Grass / Outdoor Innovation
Top comments (0)