AI Product Design - UX Research - Interaction Design
A working plant-care web app that combines AI identification, automated care schedules, and a personal plant journal into one flow.
Overview
People buy plants because they want their homes to feel more alive, natural, and personal. But care quickly becomes a repeating cycle: they get busy, forget watering or light needs, watch plants decline, then buy another plant and start over.
PlantPal breaks that loop by turning identification, reminders, and growth tracking into one care system that supports motivation after the purchase moment fades.
I conducted generative research with 6 plant enthusiasts to uncover the "tool fatigue" phenomenon. I designed a streamlined UX that consolidated 3 separate use-cases into one app.
Going beyond static mockups, I independently integrated the Gemini API to power a real identification flow, including error-recovery mechanics for AI uncertainty.
Problem
Research
Plant owners rely on multiple disconnected tools for everyday care.
The market is large, but care support remains fragmented.
Millennials identify as plant parents, turning care into a lifestyle habit.
Many houseplants die within the first few months.
6 interviews
12 survey responses
What We Heard
"I never know when to water."
"I use Notes, Google, and reminders."
"I forgot what plant this is."
Persona
They are not standalone characters; they summarize the behaviors that appeared across interviews: motivated plant owners who still forget care routines, lose plant context, and need clearer follow-through.
Alex
"I love collecting plants, but keeping up with each one's care needs becomes overwhelming."
Sarah
"I love the idea of plants, but I just can't keep them alive with my schedule."
Pain Points
The Root Cause
Users were not lacking motivation. They lacked one connected system that could move from plant identification to care guidance, reminders, and growth records.
Design Goal
ChatGPT can answer a plant question, but it does not manage the full care loop. PlantPal uses AI identification as the entry point, then turns the result into reminders, plant management, and growth records in one workflow.
The product is mobile-first because the core behaviors happen on a phone: users photograph plants, receive care reminders, check daily tasks, and update journals while standing next to the plant.
AI-powered Care Flow
01
Take photo
02
Identify
03
Schedule
04
Manage
Before
After
Designing for AI Uncertainty
Gemini recognition averaged around 70%, so the interface needed to support success and recovery instead of pretending every scan was perfect.
Design Exploration & Iteration
The Initial Concept
In my first iteration, PlantPal included four main features: AI Identification, Care Reminders, Growth Journal, and Shop.
The logic seemed sound: If users need help caring for plants, they must need supplies too. Why not let them buy soil, fertilizer, and pots directly in the app?
Version 1.0 with Shop
Four tabs competing for attention, with Shop diluting the core value proposition.
Refocused Version
Clean, focused navigation that reinforces the core workflow: Scan to Schedule to Track.
Prototype Before Testing
Before testing, I built the first interactive version with Shop included. This made the product testable as a real navigation decision, not just a theoretical feature list.
The prototype showed whether users naturally treated Shop as part of the care workflow or as a distraction from it.
In Their Words
To understand whether Shop really belonged in PlantPal, I interviewed 6 users aged 20-25 and observed 5 participants during usability testing. Across both interviews and behavior, the pattern was consistent: people came to PlantPal for plant care, identification, and reminders, not to purchase supplies.
Think
Do
Feel
Say
"I downloaded this kind of app to figure out what's wrong with my plant. If I need supplies, I already know where to buy them."
Jia Chen22 - Graduate Student
Student / Beginner Plant Owner
"The Shop tab feels extra. I'd probably ignore it and go straight to care tips or reminders."
Yuki Sato24 - Marketing Coordinator
Working Professional / Plant Hobbyist
"I already use Amazon or my local nursery. An in-app shop doesn't really add value for me."
Marcus Reed23 - IT Support Specialist
Working Professional / Plant Newbie
"I like browsing plants in person, so I wouldn't buy from the app. I just want reliable care advice."
Linh Nguyen25 - Product Design Student
Student / Plant Enthusiast
"If I saw a Shop page, I'd assume it was trying to sell me something. I'm here for plant care, not shopping."
Ethan Brooks24 - Elementary School Teacher
Working Professional / Indoor Plant Owner
"The care tools are the reason I'd keep using PlantPal. Shop makes the experience feel less focused."
Anaya Patel24 - Data Analyst
Working Professional / Balcony Gardener
What We Learned
01 x Signal
Across 5 usability tests, 0 participants clicked Shop as a key action during their core task flow.
02 x Signal
100% of participants already relied on Amazon, Target, or local nurseries for plant supplies.
03 x Signal
Users consistently prioritized diagnosis, reminders, and care guidance over commerce features.
Decision Made
Based on both interviews and usability testing, I moved away from the Shop direction so PlantPal could feel more focused, more useful, and more aligned with what users actually came to the app for.
This insight directly informed the final navigation: a tighter care loop around scanning, scheduling, and journaling.
Build Evidence
I used Figma to map the complete product system, Supabase to structure real plant data, and the live prototype to test the core care flows.
01 x Figma System
Full-screen design exploration across home, identification, garden, watering, shop, journal, and plant detail pages.
02 x Supabase
The Garden table stores plant names, image URLs, care tips, and backend records, proving the prototype is not only static UI.
03 x Early App Map
Early working files show how the app structure, plant identification screen, and backend connection were mapped before visual refinement.
04 x Interaction Draft
Interaction drafts connected implementation decisions with interface behavior, keeping the product usable beyond static mockups.
Feature Demo
This section shows the user-facing experience, separate from the work trace above: daily status, AI identification, plant collection, and growth journaling.
01 x Home
The home page turns daily plant status into a direct entry point for identifying a new plant.
02 x Identify
Users can start with a photo upload flow designed around clear scanning guidance and recovery cues.
03 x Garden
Garden cards make plant health, water status, and detail access visible without heavy navigation.
04 x Journal
The journal keeps care notes and milestones connected to the plant detail experience.
Final Solution
PlantPal is structured as a single care loop. Identification is not an isolated feature: it becomes the trigger for scheduling, daily action, and long-term growth records.
AI recognition creates the plant profile.
Care needs turn into a clear schedule.
Dashboard tasks reduce memory load.
Journal entries make progress visible.
Impact & Takeaways
The final direction is not another feature-heavy plant app. It is a connected care system where identification, reminders, and growth records support one core behavior.
Users completed the identify-to-care workflow without assistance.
The focused final flow rated highly after reducing shopping-related distraction.
Instructor commendation for execution quality and product focus.
Product Focus
Testing showed that users came to PlantPal for care support, not purchasing. The final system narrowed the experience around scan, schedule, and journal.
AI UX
Building with a real AI flow pushed the design beyond polished mockups. Confidence states, fallback paths, and fast recovery made the product feel more credible.
Reflection
What I Learned
The key design challenge was not adding AI for novelty. It was deciding where AI should enter the user journey, how much confidence to show, and how the result should become a useful care action.
Future Improvement
Next, I would improve personalization by learning from completed care tasks, missed reminders, plant health changes, and seasonal context so each plant plan becomes more responsive.
