Leafolio
A helper for houseplants that uses photos to identify, diagnose, and schedule care for everything living on the windowsill.
The spark
At some point our house filled up with plants, and I realized I couldn't keep track of what each one needed or how often to care for them. When a plant started looking sick, I'd have to dig through articles and guess what was wrong. I wanted a simple way to identify each plant, know how to care for it, and get help when something looked off.
What it is
- A system for managing houseplants that starts with a photo from your phone to identify the plant.
- Automatically creates a custom care schedule based on plant type and environment.
- A triage flow where you can upload a photo of an unhealthy plant, get a likely diagnosis, and receive a simple treatment plan.
- Follow-up reminders to check back on the plant and confirm whether it's recovering or needs a new plan.
Stack
- Photo upload and analysis via multimodal vision APIs to identify plant species and potential issues.
- A scheduling layer that tracks watering, fertilizing, and follow-up checks per plant.
- Same core web stack as other small projects (Next.js, Vercel, Supabase).
What I learned
- Working with multimodal inputs (images + text) opens up new types of everyday utilities, but it also requires careful prompt and response design.
- Building a simple scheduling system around those predictions is just as important as the model itself—people need reminders, not just diagnoses.
- Even small "niche" helpers like this are a great playground for experimenting with vision models and event-driven backends.