Case Study
Plantsy
A plant care companion app that makes keeping plants alive less intimidating.
Client
Personal project
Year
2024
Type
Mobile App, Branding
Role
End-to-end design
Deliverables
App design, Brand identity, Prototype
Most plant care apps fall into one of two categories: overwhelming data dashboards for serious gardeners, or cute but useless reminder tools. Neither solves the core problem for the "enthusiastic but forgetful" plant owner: understanding why a plant is struggling, not just being reminded to water it. Plantsy is a personal project that started with a question: what would a plant care app look like if it was designed with the same rigor as a professional health app, but with the warmth and approachability of a consumer product? I ran guerrilla interviews with 12 plant owners in two weeks. The consistent insight: people do not kill plants out of negligence. They kill them because they get contradictory advice and lose confidence.
Brand direction
The brand needed to feel warm and alive without being cute. Plants are sophisticated living things — the visual identity should reflect that. I landed on an organic typographic mark, a restricted palette of green and warm neutrals, and illustration style that felt like botanical drawings updated for a modern sensibility. The name "Plantsy" is deliberately casual — it disarms the expertise anxiety that makes plant care feel intimidating.
This was a self-directed project and that came with its own constraints: no client to push back, no one to tell me when to stop. I had to set my own scope and stick to it, which is harder than it sounds. The diagnosis flow went through four complete redesigns before I was satisfied. The breakthrough was stopping trying to make it feel like a medical app and instead making it feel like asking a knowledgeable friend. That shift — from clinical to conversational — changed every interaction detail. I shipped this project as a prototype only, not as a live product. If I were to build it, the plant identification feature would require a machine learning model that was outside the scope of what I wanted to take on alone. That's a real constraint I documented explicitly rather than hand-waving.
Next project