COMP_SCI 329: HCI Studio
Overview
LiveLeaf is an intelligent plant care companion that uses personalized, emotionally-toned notifications and social accountability to help working individuals maintain healthy plants. The project explored how adaptive messaging and community features can address forgetfulness — not through rigid automation, but by working with human behavior patterns.
The Problem
Who We Designed For
Working individuals who want plants in their homes but struggle to keep them alive. They lack the knowledge, bandwidth, or instinct to care for them consistently.
What We Found Through Interviews
Our initial needfinding used surface-level questions — "How many plants do you have?" and "How often do you water?" — which gave us scattered information but didn't reveal the core issue. When we revisited the problem with deeper, emotion-focused questions, a clearer picture emerged:
- "What motivates you to care for plants?" → Thriving plants provide a small sense of accomplishment.
- "How do you feel about not watering them?" → It's embarrassing when they look unhealthy, but easy to put off until it's obvious.
- "What challenges do you face remembering plant care?" → It's not a priority unless something triggers a reminder.
Core Insight: Plant care isn't really about logistics — it's about people feeling capable of nurturing something living. Forgetfulness stems from competing priorities, not a lack of caring.
Design Evolution
Attempt 1: The Gamified Virtual Garden
Our first hypothesis was that gamification and social pressure would motivate care. We built a virtual plant home system with care tracking, community visits, and reward mechanics.
Why it failed:
- Data entry felt tedious and became a barrier to use
- Users didn't connect emotionally with virtual plant representations — low stakes made it easy to ignore
- Shallow testing with Canva slide click-throughs didn't engage users meaningfully
- We were solving hypothetical problems, not real ones
User feedback was direct: "I don't want to type in every detail about my plant every time I water it — just keep it simple." And: "If the app nags me too much, I'll just turn it off."
The Pivot: BeReal-Inspired Social Accountability
We reframed the problem and design argument:
- Revised Problem Statement: Working individuals want to maintain healthy plants for a sense of internal fulfillment, but they struggle to prioritize care as it often slips their mind, leading to neglected plants and frustration.
- Revised Design Argument: Leveraging social pressure and personalized, toned reminders fosters motivation and accountability, addressing forgetfulness with emotionally resonant notifications.
The new solution drew from BeReal's model — push notifications prompt users to snap and share a photo of their plant, creating a lightweight social loop. Three core features:
- Personalized reminders delivered as push notifications with varying emotional tones
- Camera-first posting — snap a photo of your plant in response to the notification
- Community feed — view others' plants to build accountability and connection
Testing: Embracing Imperfection
Experimental Setup
We tested with 4 users via WhatsApp group chats, varying two independent variables:
- Feed type: Closed (post before viewing others) vs. Open (view anytime)
- Message tone: Positive/encouraging vs. Negative/urgent
Each user was assigned a different combination and received toned reminders over multiple days.
Challenges We Faced
Our testing process was messy — and we learned as much from that mess as from the results:
- Not enough users to draw statistically significant conclusions
- Changed our experimental setup too many times — we kept revising the number of groups, control variables, IVs, and DVs
- Started with one group chat for all users, then pivoted to individual chats mid-experiment
- Spent too long debating the perfect experiment instead of just getting actionable feedback
Results
User-by-User Findings
User 1 (Closed Feed, Positive Messages)
Was so anxious about posting the "perfect" plant photo that they kept pushing it off and never posted. Social pressure backfired — instead of motivating action, it created performance anxiety.
User 2 (Open Feed, Negative Messages)
Was away from home and physically unable to water their plant. The urgent-toned messages had no effect. This highlighted a need for context-awareness — an "away from home" feature.
User 3 (Closed Feed, Negative Messages)
Engaged consistently but reported mixed feelings: helpful some days, tedious on others.
User 4 (Open Feed, Positive Messages)
Engaged actively and even asked why we stopped sending messages — a signal of genuine value.
Key Takeaways
- Peer pressure is a double-edged sword. For some users, social posting motivated care. For others, it created anxiety that prevented action.
- Tone variation and feed type showed no clear difference with our sample size — more users are needed.
- Context matters enormously. The same notification can be helpful or irrelevant depending on the user's situation.
- Users 3 and 4 posted consistently but with different levels of enjoyment, suggesting personality and intrinsic motivation play a large role.
Reflections & Lessons Learned
On the Design Process
- Watch for patterns in user needs rather than jumping to solutions
- Don't over-engineer — we tried to solve too many problems at once with the gamified garden
- Going "backwards" is okay — returning to needfinding after our first prototype failed was the right call
- Progress over perfection — shipping an imperfect test taught us more than endlessly refining the plan
On AI-Assisted Plant Care
- Successful technology must work with human behavior, not against it
- Users want gentle accountability without feeling judged
- Adaptive, context-aware messaging is more effective than fixed reminders
- Transparency in why a notification is sent builds trust
Next Steps
- Add an "away from home" feature so notifications pause when users can't act on them
- Conduct follow-up interviews to understand why some users enjoy the app more than others
- Expand the user pool to generate more meaningful data on tone and feed-type effects
- Explore predictive scheduling — learning when users are most likely to act and timing notifications accordingly
- Investigate plant health assessment via photo analysis to provide actionable, personalized care guidance