Valentine's flowers die. This one doesn't.
The three of us were swapping Valentine's Day stories when we realised we'd all done the same thing, bought flowers, watched them wilt in a week, thrown them out. It felt wasteful, but more than that it felt like a missed opportunity. The gesture disappears with the flowers.
So we asked: what if the gift itself did the emotional heavy lifting and stuck around? We locked in three non-negotiables, permanent so the recipient keeps it, personal so it feels made for them specifically, and interactive so the moment of receiving it is something they actively participate in, not just observe. Those three constraints shaped every decision that followed.
Personalisation mattered to us beyond just the concept. A generic "Will you be my Valentine?" wasn't enough, we wanted the giver to write something only their partner would recognise, a message that meant something to the two of them specifically. Same thinking applied to scent: different people respond to different fragrances, so the mist diffuser was designed to work with whatever the giver chose to add. The idea was that two people could receive the exact same bouquet and have it feel completely different, because it was built around them.
Meet Alex K., and why the persona actually mattered
Early on we realised we were at risk of designing for ourselves, three people who are comfortable with electronics and think hologram bouquets are obviously cool. So we built a persona to keep us honest.
Alex is the gift giver. Wants to make someone feel special but gets anxious about execution. Hates things that feel generic or require a manual to operate. In our mock interviews, the same fear surfaced repeatedly:
"I dont know how to use it without instructions"
That single insight drove two of our biggest decisions: why we moved away from a push-button to a touch sensor (more intuitive, less "press here"), and why we eventually added a 3D-printed tag pointing to the sensor (because even intuitive isn't obvious enough when emotions are running high).
An Arduino, some tulips, and a Victorian magic trick
Every component was a deliberate choice, not just a technical one.
We chose a capacitive touch sensor over a push button because touch feels intimate, pressing a button feels like operating a device, touching a flower feels like interacting with a gift. The RGB LEDs glow red from inside the bouquet because the light needed to feel like it was coming from within the flowers, not from a separate screen or gadget bolted on.
And the LCD + Pepper's Ghost illusion, this was the most deliberate choice of all. We could have just shown the message on a screen. But a screen sitting in a bouquet breaks the magic; it reminds you there's a device in there. Pepper's Ghost is a 19th century stage trick that uses angled glass to make a reflection appear to float in mid-air. It meant the message could emerge from the flowers themselves, not from a display. The reveal feels impossible, which is exactly the point.
Three builds. Each one broke something we thought we'd solved.
We deliberately tested at 50% before adding any aesthetic finishing. The reason: we needed to know if the concept worked before investing time in making it look good. If people didn't respond emotionally to the idea, no amount of nice wrapping would fix it.
We were right to check early. Everyone said they'd feel genuinely valued receiving something like this, the concept landed even with exposed breadboard and dangling wires. But the feedback was almost entirely emotional and very light on usability critique, because there wasn't enough finished product to critique yet. That was fine. It told us what we needed to know: keep going.
Adding the flowers changed everything about how people related to the object. Suddenly it looked like a gift, and people tried to actually use it, which immediately exposed the sensor discoverability problem we'd underestimated. Nobody touched the sensor without prompting. Some walked right past it. This is what pushed us to design the 3D-printed instructional tag: not a failure of the sensor, but a failure of affordance.
The scent feedback was mixed, some loved it, others were indifferent or put off, which raised the question of whether we should add real fragrance or leave the diffuser scent-free. Then came the moment that changed our thinking entirely.
The mist diffuser at this stage was running plain water. No scent at all. But because mist was rising around flowers, multiple participants commented on the rose fragrance. Their brains filled in a scent that simply wasn't there, the visual context of flowers and mist was enough to trigger a phantom smell.
The 75% prototype still had one unsolved problem: the message display. An LCD screen sitting among flowers looked exactly like what it was, electronics attached to a gift. It broke the illusion every time. We needed the message to feel like it belonged in the bouquet, not bolted onto it.
Pepper's Ghost solved this. By angling a piece of glass above the LCD, the reflection of the message appears to float above the flowers with no visible source. We also added the 3D-printed tag directly in response to the 75% discoverability feedback, and wrapped the whole thing properly so the first experience was receiving a gift, not inspecting a prototype.
The full experience
Watch for the hologram moment, that's the one. Everything before it is build-up.
Everyone loved it. Almost nobody knew how to use it.
Final round had 6 participants, rated on thoughtfulness 1–5. The scores skewed 4–5 across the board, emotionally it landed. But two friction points came up consistently enough that we couldn't ignore them.
First: sensor discoverability. Even with the tag, people weren't always sure what to do first. The interaction flow wasn't obvious enough without prior context. Second: the message was mirrored on the LCD, readable if you tried, but disorienting on first look. Both pointed to the same underlying issue: we'd optimised for the reveal and underinvested in the journey to get there.
What we'd do differently
- ♡Lead with the journey, not the reveal. We spent so much energy on the hologram moment that the path to it, finding the sensor, understanding what to do, was underdeveloped. A motion sensor that triggers automatically as someone picks up the bouquet would remove that friction entirely.
- ♡Design the environment, not just the object. The phantom scent proved that context does as much perceptual work as components. We'd lean into that more deliberately next time, think about lighting, setting, even the wrapping as part of the experience.
- ♡Test the concept before you build anything. Our 50% test validated that the emotional core worked before we'd invested in fabrication. That's a habit worth keeping, the feeling should be right before the form is.