H2Know
I designed a plant-care prototype with an Arduino soil-moisture sensor and a companion app for checking plant health.
ToolsFigma · Arduino · Illustrator
The four strongest screens
Jump to 05, the app
Jump to 05, the app
Jump to 04, the hardware
Jump to 05, the app
A plant can't tell you it needs water.
It sits out of your sightline, and it can't tell you anything until it's already dying.
The sensors people buy don't help beginners.
A sensor comparison, then three interviews.
Sage, deep teal, warm yellow.
An earthy palette, set in Spoof Medium.
Five screens, one sensor.
A five-screen app concept beside an Arduino prototype.
What I learned.
- The naming mechanic needed more time.
- Two features didn't make the build.
Role, process and impact
Role, process, impact and methods
- Role
- I led the research and designed the app, sensor interaction, and hardware prototype.
- Process
- I interviewed three plant owners and used their care routines to design plant-health feedback, naming, and rewards.
- Impact
- I developed a five-screen app concept alongside an Arduino soil-moisture prototype.
- Role
- Solo Designer
- Timeline
- 6 weeks · 2024
- Methods
- User Interviews · Prototyping · IoT
- Tags
01Problem
A plant can't tell you it needs water.
- A plant sits in a corner, out of your sightline, and can't tell you anything until it's already dying.
- Smart pots cost $60–120, moisture meters spit out numbers you can't read, and reminder apps nag you like every other notification.
- You water a plant because you care whether it lives.
Read more
Last year, Americans spent over $48 billion on indoor plants. Yet 67% of plant parents admit to killing at least three plants in the past year (Live Science, 17 April 2007). The cause is distance. A plant sits in a corner, behind a desk, out of your sightline, and it can't tell you anything until it's already dying.
The products meant to fix this fall short. Smart pots cost $60–120, basic moisture meters spit out numbers you can't read, and reminder apps nag you like every other notification. You don't water a plant because you're missing information; you water it because you care whether it lives.
“What if your plants could actually tell you exactly what they need, when they need it?”
The design question that started the project
“I'm happy when they die because then I can buy a new one. Every time I'm depressed I get a new one.”
Anya, 16 · Interview participant
HOW MIGHT WE
- Motivate users to water plants in a non-invasive, helpful way?
- Educate users about their specific plants without overwhelming them?
- Create a straightforward, familiar interface that requires no learning curve?
- Create a stronger emotional bond between owners and their plants?
02Research
The sensors people buy don't help beginners.
- I compared the sensors people already buy: readings are hard to interpret, calibration drifts, and nothing reminds you.
- Then I interviewed three plant owners.
- Their plants died because the relationship had gone cold.
Read more
I compared the sensors people already buy (the Gikfun capacitive sensor, basic moisture meters) and watched them fail beginners: readings are hard to interpret, calibration drifts, and nothing reminds you or makes you care. Then I interviewed three plant owners. Each told a version of the same story: their plants didn't die because they didn't know how to care for them. They died because the relationship had gone cold.
What they wanted
- Clear, low-effort reminders that don't feel like nagging
- Feedback that tells them what to do, not just sensor numbers
- A reason to check in (emotional connection, not obligation)
- Forgiveness when they forget (hardy plants, low-maintenance species)
What pushed them out
- Out of sight, out of mind (plants in corners, behind furniture)
- No feedback until death was already imminent
- Overwatering as often as underwatering, no confidence in timing
- Existing apps felt like homework, not companionship
03Design logic
Three findings, three features.
- One design target: build a relationship with the plant.
- Guilt makes absence visible, ownership gives each plant a name, a species and a birthday, and reward makes care pay off.
- When the soil goes dry, the LED, the app icon and the Friends feed all flag you at once.
Read more
The research gave me one design target: build a relationship, not a dashboard. The interviews kept circling three psychological levers, and I made each one drive a part of the app.
- Guilt. Make absence visible. When the soil goes dry, the LED turns on, the plant's face in the app icon drops from happy to sad, and the Friends feed tells everyone: "Pete is 9 days late on watering Todd: remind him now." The hardware, the interface, and your friends all flag you at once.
- Ownership. Give it a name, a species, a birthday. Each plant has a name (Jeremy, Todd, Bill), a species (Golden Pothos, Mini Monstera, Scarlet Ball Cactus), and an adoption date ("Your plant since June 20, 2023"). The detail screen tracks its age in years, days until the next watering, and days until the next fertilize. Once a plant has a name and a history, you stop letting it slip your mind.
- Reward. Make care pay off. Points accumulate with each watering. Hit a threshold: redeem real coupons: "15% off fertilizer at Home Depot." Badges mark milestones: "Best Friend" (owned for one year), "Super star" (20-time watering streak). The friends feed also runs positive: "Robert just watered Finn: congratulate him on his 4 time streak!"
04Hardware
Hardware + app as one system.
- An Arduino Nano with a capacitive soil moisture sensor detects dry soil and signals with an LED.
- It pairs with a five-screen companion app for plant health, the social layer and real-world rewards.
- The whole physical layer is a kit of about $15.
Read more
H2Know is a two-part system: an Arduino Nano with a capacitive soil moisture sensor that detects dry soil and signals via LED, paired with a five-screen companion app that visualizes plant health, handles the social layer, and converts consistent care into real-world rewards.
3Plant owners interviewed: Anya, Macey and Pete
5Screens: Home, Water Today, Plant Detail, Friends, Coupons. Each built around one motivation lever.
67%Of plant parents admit to killing at least three plants in the past year (Live Science), the failure H2Know targets with guilt, ownership, and social accountability
$15Approximate cost of the Arduino hardware kit, the entire physical layer for a market where "smart" planters retail at $60–120
05App
Five screens, each built around one motivation lever.
- Home lists each plant with a face that goes from happy to sad with the sensor data.
- Plant Detail tracks age, watering and fertilizer, with awards; Water Today splits "To water" from "Completed".
- Friends flags late watering and congratulates streaks, and points unlock real coupons.
Read more
- Home
- Your plant list: each plant shown with its emoji avatar face, name, and species. The face state (happy / neutral / sad) updates based on sensor data. Adding a new plant opens a 5-minute survey that recommends the right species for your space and habits.
- Water Today
- Two-column view: "To water" (Todd, 100 ml) next to "Completed" (Jeremy, checkmark). You read the day's task in one glance, without digging through a list.
- Plant detail
- Age, days until watering, days until fertilize, health status bar, weekly watering chart by day. Awards live here: "Best Friend" (one year), "Super star" (20-streak). Plant birthday shown at bottom.
- Friends
- Social accountability feed. Late watering generates a public flag and a "Remind him now" prompt. Streaks generate congratulation prompts. Both directions (shame and celebration) come from the same screen.
- Coupons
- Points accumulate per watering. At threshold, real coupons unlock: "15% off fertilizer at Home Depot." Progress bar shows points to next reward. Redeemed coupons stay visible as record of care.
06Reflection
The naming mechanic needed more time.
The full story, in order, lives in the deck.
View presentation PDF ↗The naming mechanism deserved more design time.
Every interview pointed the same way: named plants outlive unnamed ones. Yet in the prototype I made naming a plain text field on the setup screen. If I framed it as a small ritual instead, with onboarding that treats the name as the start of a bond, the ownership lever would kick in from the very first tap.
Hardware and software need to be co-designed.
I built the Arduino prototype and the Figma app in separate phases. So I designed the hardware without knowing which app interactions it would trigger, and designed the app without knowing how slow the sensor actually was. Next time I'd prototype the physical and digital sides side by side from week one.
Two features didn't make the build; both would have changed the testing.
I cut two features from the prototype: a plant locator map (which plant in the house needs water, based on sensor placement) and a propagation gifting system (trading clippings with other owners). Both go straight at the out-of-sight problem. The locator surfaces the plants you've stopped seeing, and gifting raises the social stakes of keeping yours alive. I'd build both in v2.