Live app
Plant ID App
Why it’s useful
During my time off work, I have spent a lot of time in the garden. I wanted an app I could use every day to help me tend my plants.
What it does
- Accepts an uploaded plant photo.
- Uses AI vision to identify the plant and shape the result into care guidance.
- Provides practical notes on watering, light, placement, and pet safety.
Current status / Next steps
Current status
- Live app at plants.gavinnesom.com.
- The app accepts plant photos and returns a structured care report.
- Server-side rate limiting is in place before the OpenAI vision API call to protect my personal API usage.
Next steps
- Improve the care-report formatting.
- Add stronger uncertainty and safety language where identification confidence is limited.
- Add export to PDF.
Build / Deployment
Vite · React · JavaScript · Tailwind · Vercel · Upstash Redis · OpenAI vision API
The Plant ID App is a standalone Vite/React web app. The browser handles photo upload and report display, while a Vercel serverless API route applies rate limits before calling the OpenAI vision API.
Front end
Vite, React, and Tailwind provide the upload flow and care-report UI.
Serverless API
A Vercel API route keeps the OpenAI vision API call off the client and applies rate limiting before the request is sent.
Rate limiting
Upstash Redis stores request counts so the public demo cannot freely burn through my personal OpenAI API usage.
Architecture / How it works
Upload flow
The browser accepts the plant photo and shows the generated care report.
Vision request
The serverless API route sends the image request to the OpenAI vision API without exposing the API key in the browser.
Rate limiting
Upstash Redis tracks usage before the OpenAI call is made.
Care report
The response is shaped into identification, watering, suitability, safety, and next-step notes.
Input safety
Uploads are checked by file type and size before processing.
Design Principles
Useful, not just identified
- A plant name is only the starting point; the result should help someone decide how to water it, where to place it, and whether it belongs in their home or garden.
- The care report should read like practical garden help, not a raw AI response.
Safety and uncertainty
- Plant identification is not perfect, so confidence, caveats, and safety notes need to be visible.
- Pet-safety warnings matter because plant advice can affect real choices around a house, yard, or garden.
Readable care cards
- Customized icons make watering, light, suitability, safety, and care notes easier to scan.
- The page should make the important care categories obvious at a glance.