← Work · Interactive data story

IUNU × Bushel Boy

An interactive case study for the greenhouse-vision company IUNU and its grower, Bushel Boy Farms — turning dense crop data into a human story about trusting what the plants are telling you.

IUNU × Bushel Boy interactive case study hero, over greenhouse tomatoes
Optimizing greenhouse production with automated crop registration, powered by IUNU's LUNA AI.
Context
Bushel Boy Farms — a leading Minnesota & Iowa greenhouse grower — needed to show how full-coverage crop data changes the way growers steer a crop.
Role
Interactive storytelling, design, and front-end — turning a dense report into something you explore.
Built with
Semantic HTML · CSS · vanilla JS · scroll-driven interactions

The challenge

To steer a crop, growers pair expert observation with weekly crop registration — measurements taken by hand. At Bushel Boy, that meant sampling under 0.2% of all plants, once a week.

  • Labor intensive. Data was gathered by hand, one day a week.
  • Blind to anomalies. Anything outside the tiny sample went unseen.
  • Hard to trust. Manual sampling carries human bias and inconsistency.
“Crop registration is done manually, and when we rely on people to take this data it is not always accurate and includes bias. It's hard to trust the data.” Felix Tarrats · Horticultural Consultant, Bushel Boy Farms
Interior of a Bushel Boy greenhouse with rows of tomato plants
Bushel Boy's greenhouses produce up to 30× more, with less water, than outdoor operations.

The signal

Growers already use head-to-flowering-truss height to judge whether a crop is running too vegetative or too generative. With IUNU's LUNA AI providing full weekly coverage instead of a hand sample, that measurement became a reliable production signal.

The real gain is timing. A weekly hand-sample means you react after the crop has already changed; full-coverage data lets the team act on climate strategy before problems compound.

Weekly signal

Head to flowering truss

Weekly head-to-flowering-truss height, weeks 28 through 31 The measurement falls from 16.6 centimetres, above the target range, to 13.0, 11.3, and 11.9 centimetres within the 10 to 15 centimetre target range. Target · 10–15 cm 5 10 15 20 16.613.011.311.9 W28W29W30W31
  1. Week 28 · 16.6 cmToo vegetative · 22.0°C inside · 20.0°C outside · 14,000 J light
  2. Week 29 · 13.0 cmTarget · 22.4°C inside · 21.0°C outside · 15,800 J light
  3. Week 30 · 11.3 cmTarget · 25.7°C inside · 24.0°C outside · 15,700 J light
  4. Week 31 · 11.9 cmTarget · 24.2°C inside · 23.0°C outside · 13,000 J light

After irrigation changes, the crop moved from too vegetative in week 28 back into the target range by week 29 and remained there through week 31.

From samples to spatial confidence

The story's turn is a shift in coverage. A before/after view contrasts isolated weekly hand-samples with a greenhouse-wide picture of crop balance, so anomalies that used to hide between samples become visible.

From there, truss height sorts every bay into three balance states — the same colors as the chart — so a grower can read the whole greenhouse at a glance and steer with confidence.

“Plants don't lie, and when you see the data from the plants you know exactly what is going on.” Roberto Ramirez · Head Grower, Bushel Boy Farms

Before · manual

<0.2% sampled

Isolated hand measurements left the space between samples unseen.

After · LUNA AI

500× coverage

Greenhouse-wide registration surfaced balance and anomalies across the crop.

Bay balance

Read the crop, then steer it

Week 28 1 of 12 bays in target

  1. 16.6
  2. 16.7
  3. 19.6
  4. 17.3
  5. 17.5
  6. 17.8
  7. 13.7
  8. 15.7
  9. 16.3
  10. 16.7
  11. 15.7
  12. 15.1

Week 31 11 of 12 bays in target

  1. 11.7
  2. 10.7
  3. 13.3
  4. 13.2
  5. 11.5
  6. 12.1
  7. 12.4
  8. 12.3
  9. 12.8
  10. 11.5
  11. 11.5
  12. 9.9

Too vegetative >15 cmTarget 10–15 cmToo generative <10 cm

Storytelling decisions

The source material was a dense case-study report. My job was to make it felt: lead with the grower's own words, let a reader step through the weekly signal themselves, and reveal the before/after coverage as an interaction rather than a claim. Each section earns the next — problem, signal, coverage, action — so the payoff (trusting the data) lands as something you discovered, not something you were told.

  • Interactive storytelling
  • Data visualization
  • Front-end
  • Scroll interactions
  • Art direction

Open for new projects

Got a creative idea?
Let’s give it a pulse.