agsight
jan 2024 - present
helping 24 specialty crop farmers grow more equitably
A hand wrapped in a produce wreath holds a screen with Agsight's plot overview screen.
30-second synopsis
I created a machine learning app to help 24 vineyards and orchards in 2 countries respond to crop threats without sensors, improving average yields by ~12%. Helped 4 winegrowers reduce smoke taint during the 2025 California wildfires and 5 citrus/avocado farms overcome saltwater intrusion. Using this work, I successfully advocated for a new agribusiness course to expand agricultural education in Frisco ISD by petitioning the superintendent and School Board.
24

farms impacted

12%

improvement in yield

36%

decrease in crop stress

98%

as accurate as sensors

01
auditing the current experience
analyzing the problem
To uncover common problems, pain points, and industry gaps, I surveyed 150+ specialty crop farmers across the US using agtech over their daily challenges, motivations, resources, and habits. I distilled responses into 6 key threads.
A scatter plot shows responses for "What has been your return on investment for your agtech solution?" The graph shows an increasing linear trend between yield increase and the price of agtech solutions: As the price increases, yield also increases, but marginally.
Farmers were frustrated by the unclear ROI of agtech tools. Many found it challenging to justify the high costs when the outcomes were not immediately evident. One shared that after spending $5K on a precision farming system, their yield only increased by 1%.
Bar chart comparing agtech solution effectiveness across crop types, showing common staples as most supported, heirloom vegetables as moderately supported, and fruits as least effectively supported, with varying levels of perceived ineffectiveness and detriment.
Existing tools' lack of location-specific personalization result in data that were not only ineffective but sometimes detrimental. 34% of surveyed farmers reported that agtech solutions frequently overlook the needs of specialty crop farmers, including those growing heirlooms or fruits.
A funnel chart shows the percentage of surveyed farmers who were able to successfully schedule a demo, acquire infrastructure, complete setup, training, and start using an agtech solution. Only 54% of those who scheduled a demo started using the agtech solution.
Many farmers reported that onboarding for many agtech solutions is lengthy and complex because it takes several weeks to schedule demos, acquire infrastructure (notably sensors), and finish training sessions. 83% of farmers over 50 reported struggling with a steep learning curve.
A pie chart illustrates responses to the question "What is the primary challenge you face when using your current agtech solution?" with 43% citing irrelevant features, 31% hard to interpret data, 14% limited applicability, and 4% no major issues.
Many farmers complained about data being presented in cluttered interfaces in a spreadsheet format, which made it challenging to quickly find specific details or parse through complex data. 43% of farmers using agtech software also reported that their current solution had too many features that were not relevant to their specific needs or crops.
A stacked bar chart shows responses to the the question "How often do you find that the data provided by your agtech solution lacks actionable insights?" across four categories: Weather monitoring, soil monitoring, crop health, and pest detection. The majority of farmers report "often" or "always".
Farmers reported that most data provided by agtech solutions lacked actionable, farm-specific insights, which made it difficult for them to use in real-time decisions. The data was presented in a vacuum, not applied for recommendations.
Bar chart showing that most users spend $1K–$2K per year on agtech tools, followed by $251–$1K, $0–$250, $2K–$5K, and the fewest spending over $5K.
91% of surveyed farmers cited the high price of agtech solutions as a major barrier to adoption. Primarily, high upfront costs and recurring fees put them out of reach for small- to medium-sized farms that typically operate on tight margins, which disincentivizes them from using agtech.
02
Aasking the right people the right questions
researching my audience
I read up on industry trends, interviewed farmers, and analyzed competitors to spot market gaps and set benchmarks. By working closely with farmers, I learned what they truly want, how they consume content, and where existing tools fall short—including the moments that cause them to drop off.
8

interviews

153

survey responses

15

competitive audits

The user journeys of Jessica, Henry, and Michael.
I mapped each user archetype to their journey on competitor platforms by tracking their goals and outcomes. This helped me spot pain points, uncover opportunities to improve, and identify which features to prioritize.
Competitive audit sheets highlighting competitors' strengths.
I analyzed 15 top agtech tools by comparing features, user feedback, and gaps to find opportunities for innovation and differentiation. One key insight: the median competitor price was around $750/year.
A spreadsheet evaluates and categorizes the sentiments of App and Play Store reviews of Agsight's competitors.
Farmers frequently express frustration with high costs, inaccurate data, and friction during onboarding, while others expressed delight on unique features like smoke taint analyses and a focus on small producers.
A series of interviews and meetings with farmers via Zoom.
I conducted 8 virtual interviews via Zoom and chose to target Central CA specialty crop farmers of small- to medium-sized farms because they were generally most affected by the inequities posed by current agtech solutions.
defining my audience
I dove deep into the behavioral data of farmers to understand them better by defining 3 archetypes representative of specialty crop farmers and mapped them to their respective jobs-to-be-done.
An early-stage grower building a specialty crop farm with a sustainability mindset, interested in tech but limited by capital and unsure where to start.
challenges
  • Overwhelmed by the number of agtech options, most of which seem geared toward large-scale operations.
  • Wants tools that grow with their business, but many products lock them into high upfront costs or complex infrastructure.
Nneeds
  • A flexible solution with a low barrier to entry: something affordable now, but capable of expanding.
  • Clear evidence of how insights improve yield, reduce waste, or drive ROI, without requiring sensors, drones, or complicated tech stacks.
A black farmer balancing a peach on her bucket hat.
A long-time specialty crop farmer managing a small, family-run operation, increasingly pressured to adopt agtech despite limited familiarity with it.
challenges
  • Finds most agtech overwhelming due to overly technical UIs and jargon that don't align with how they’re used to working.
  • Struggles to justify investing in expensive hardware or subscriptions when the ROI isn’t immediately clear or measurable.
needs
  • An intuitive solution that requires no new hardware and provides clear, actionable recommendations.
  • Support materials or onboarding for non-technical users—something they can learn without hours of training.
A farmer stands in his orchard with a grocery tote bag full of fresh produce.
A mid-size farm manager responsible for field operations and administrative decisions juggling production and labor with minimal support staff.
challenges
  • Don’t have the time to navigate dashboards filled with irrelevant metrics for crops they don’t grow or conditions that don’t apply.
  • The steep learning curve and lack of crop-specific insights lead to underuse or abandonment within a few weeks.
needs
  • A tool that delivers local, crop-specific recommendations in a clean format so they can make quick decisions on the field.
  • A platform that reduces their mental load, with minimal setup, no required hardware, and instant visibility into what matters most: irrigation, stresses, and yield.
An elderly farmer peers into his fields behind his tractor.
03
synthesizing research
insights + opportunities
I distilled my research findings into 7 key insights.
eliminate the need for infrastructure
Smaller farms simply can’t afford to install dozens of sensors—yet that’s been the only way to get detailed insights. These sensors are pricey, fragile, and often don’t survive the season. Although sensor-free farming has been discussed in scientific literature, it’s never made it to the real world.
Here, I decided to do something radical: Opt for a sensor-optional solution.
make KPIs actionable
Across interviews and surveys, I kept hearing the same thing: farmers could access plenty of metrics like evapotranspiration and pest pressure—but didn’t know what to do with them. Numbers were everywhere, but guidance was nowhere. The real need? Clear, actionable advice tailored to their fields, like “adjust irrigation by 20 minutes” or “apply soil treatment here.” Whoever bridges the gap between data and decisions with simple, trusted steps will earn farmers’ trust and adoption.
make solutions affordable
In every interview, cost came up. One farmer told me she skipped trying an agtech tool because the subscription would eat into her fertilizer budget for the season. Another couldn’t afford repairs when their sensor system broke mid-harvest. For small- to mid-sized farms on razor-thin margins, this is normal. The opportunity? Build a solution that respects those limits—no hidden fees, no costly hardware, and clear value from day one.
personalize to improve relevance
With 47% of the surveyed farmers citing the high cost of agtech as a major barrier to adoption and 55% showing a low willingness to pay due to lack of location-specific personalization, there's a lack of personalized solutions that adapt to individual farm conditions and crop types without extensive consultations.
reduce friction dramatically
One farmer told me it took him 3 weeks just to schedule a demo, get his sensors delivered, and sit through a 2-hour onboarding call. Every extra step, login, or setup screen becomes a reason to walk away during harvest season when time is scarce. The opportunity is in building something that works out of the box—no hardware, no training manuals, no delays. The fewer the steps, the more farmers will use it—and see results.
smoothen the learning curve
When I asked farmers what kept them from using the agtech tools they’d already paid for, most talked about feeling overwhelmed. One grower, 68, said he gave up on a dashboard after three logins because he couldn’t figure out how to set up his field zones. Another said the interface felt like “Excel threw up.” With 32% of the surveyed farmers over 65 and only 35% holding a college degree, the opportunity is in removing friction. That means zero-setup onboarding, interfaces that show only what’s needed, and agtech that doesn’t assume a background in data science by meeting farmers where they are instead of expecting them to catch up.
automate manual tasks
Many farmers told me the same thing in different words: they’re exhausted by error-prone manual tasks — checking soil moisture, walking rows to spot early signs of stress, tracking pests on clipboards, or calculating irrigation needs. As such, the opportunity is in automation that flags what matters and frees farmers to focus on higher-level decisions.
04
forming a business
from problems to solutions
I used this time to ideate multiple solutions to discover ideas quickly. My focus was to diverge first, converge later. Eventually, I landed at a SaaS business model offered as a cost-effective, sensor-free, spatial machine learning Android® app to help specialty crop farmers combat vegetation stresses, water scarcity, and soil infertility without sensors.

data in a vacuum

incompatible infrastructure

steep learning curve

no local personalization

reliance on manual labor

unsustainable yield

unaffordable solutions

predicting pests + diseases

focussing the solution
Drawing from user insights and ideation sessions, I brainstormed both must-have and bold feature ideas—each tied to six core experience pillars. With the vision and scope clear, I jumped into design and engineering.
home
  • fields
  • yields
  • notifications
  • Sspatial agendas
crop growth
  • crop assessments
  • personalized schedule
  • phenology
  • crop rotation
crop threats
  • crop stresses
  • diseases
  • IPM
  • climate threats
water use
  • water needs
  • water/salinity stress
  • irrigation
  • soil moisture
soil fertility
  • health
  • nutrition
  • fertilizer use
  • composting
other
  • IoT automation
  • climate control
  • nutrient dosing
  • hydroponics
receiving deedback
Around this time, I had the opportunity to present my idea at the McFerrin Center for Entrepreneurship to deliver 2 presentations to faculty and industry professionals and secure seed funding as part of the 2024 Texas A&M Ideas Challenge, through which I acted on some (painful) but much needed feedback.
A collage showing a science fair-style project display titled "Agsight," handwritten notes with feedback on the project, and a worksheet outlining project resources, goals, and content.
what I learned from feedback
My approach tackled too many problems (agronomic, economic, geographic, etc.), which risked being too broad to effectively validate or implement. As such, I narrowed the focus to a single crop or issue within a defined geographic area, in this case by helping smallholders (orchards, vineyards, specialty crop farmers) primarily in central California improve their yield by combating vegetation stresses related to crop growth, water use, and soil fertility.
A phone perched on a rock and leaning on an avocado showing a screen with tips and best practices for raising avocado. An avocado branch balances above the tip of the top left corner of the 25-degree tilted phone.

optimize yield with spatial insights.

save big on water: irrigation and stress.

enhance soil fertility and nutrients.

diagnose any disease or pest in seconds.

AI-mazing ways to sustain yield
When developing, my biggest priority was speed-to-decision. So I built Agsight's AI pipeline to process phenology, growth, and local climate data in tandem and suggest exact next steps to diagnose which disease or pest; understand where and why it was spreading; and recommend what to do now.
diagnose and treat any plant disease or pest in seconds.

threats

track your crops' growth with beautifully precise data.

phenology

get instant updates so your crops get attention when they need it.

health

parse through location-specific information with ease.

localization

automating irrigation and climate control
Given the depth of personalization that Agsight's existing AI features had among our trial farms, I decided to extend them by allowing farmers who had access to IoT (e.g., sensors) to completely automate irrigation and climate control — both indoors and outdoors — more precisely than ever before.
agsight works indoors, too.
By tracking factors such as temperature and humidity, Agsight helps optimize light conditions for each growth stage to reduce crop stress and lower the risk of disease.
A mobile app screen shows lighting data for Shelf B1 in an indoor farm, with hotspot markers on lettuce crops and a graph tracking chlorophyll concentration over time.
irrigation, automated.
Agsight automates irrigation by using real-time soil moisture data to deliver precise, crop-specific amounts of water, personalized to growth stage and plant type.
A mobile app screen displays collected data from Row B in a vertical farm and shows live camera footage of lettuce and basil under grow lights, moisture sensor status, and irrigation controls.
rich in nutrients, rich in sustainability.
Agsight supports hydroponic systems by tracking changes in water nutrient profiles and adjusting inputs based on the specific needs of each crop. This precision reduces reliance on chemical fertilizers and can increase yields by up to 25% in fast-growing greens like kale, basil, and arugula within weeks.
A mobile app screen displays a lettuce growth overview with a projected harvest time in late June and a line graph tracking electrical conductivity levels from April to August.
ideal lighting conditions in seconds.
Agsight’s AI scanner maps light distribution across grow zones to detect underlit or overexposed areas in real time by recommending simple fixes like repositioning LEDs or adjusting intensity.
diagnoses, lightning fast
One of the most prominent uses of our machine learning models was to identify, classify, track, forecast, and manage plant diseases, pest infestations, vegetation stress, and pesticide use at the tap of a button. With an average accuracy of ~98%, it provides details on stress signals, personalized treatment, pesticide suggestions, and predicted development.
Agsight diagnosing apple scab.
powerful protection ahead of time
Farmers can assesses threats from air quality and climate to soil fertility and pests, right at their fingertips. Wildfires, frost, heat waves, pest threats — they can know ahead of time.
Agsight's notification screen showing the most urgent priorities, including a wildfire smoke taint risk.
effortless irrigation
Many farmers told me they wasted time and water because they had no way of knowing which sections of their fields actually needed moisture as weather and crop needs changed week to week. So I designed a system where growers let Agsight interpret that data by analyzing seasonal patterns, crop type, and weather forecasts. Instead of sticking to static irrigation plans, farmers can now get real-time, field-specific recommendations.
A mobile app screen displays a water assessment for apple crops, showing a recommended irrigation level of 1.1 inches per week and a disease alert warning about apple scab.
apply water where your crops most need it.

water needs

know where, when, and how to irrigate your crops and plots.

irrigation

high-quality soil results, in seconds
When I asked growers how they decide where to fertilize or whether their compost is “ready,” the answers ranged from “I eyeball it” to “I wait two weeks for a lab report and hope it’s still relevant.” That gap drove me to build a workflow where a single photo of soil becomes a full nutrient and texture profile within minutes.
personalized schedules
I built Agsight’s task engine to cross-reference crop health, weather forecasts, and phenology, then create a prioritized daily to-do list (delivered as a notification and spatial map) so farmers always know what to act on first, and where.
Three mobile app screens display a farming agenda interface showing daily task suggestions, wildfire smoke risk alerts, and a satellite map with prioritized plot-specific schedules.
06
impacting farms, impacting people
deployment
After I developed my MVP, I initiated outreach to 75+ farms, 8 of which initially agreed to trial Agsight. Currently, I've expanded this pilot program to 24 orchards, vineyards, and other specialty crop farms, primarily in California and Texas.
The Agsight website homepage promotes an AI farming app with a photo of a hand holding a phone displaying the app, surrounded by vegetables worn as bracelets, and text offering a $12.99/month subscription after a 30-day free trial.
I focused on clarity and trust by making sure farmers could immediately understand what Agsight does and why it matters to their daily work.
Joon stands alone on stage under a spotlight in front of closed blue curtains, presenting to an audience in a dimly lit auditorium.
In the audience: Superintendent Dr. Mike Waldrip, School Board members, and 150+ students and teachers. Using Agsight to promote agricultural education in Frisco ISD high schools.
A hand uses a gold kitchen sprayer to rinse fresh vegetables including kale, tomatoes, and lemons in a marble sink.
Helping 5 California specialty crop farmers overcome saltwater intrusion to increase their yield by an average of ~10%.
A black and white photo shows a window with curtains and a small balcony adorned with potted plants and flowers, partially framed by a leafy tree.
Improving agricultural equity for local businesses in Seoul by using metro farms to consolidate commodity chains between farms and restaurants.
A collage of students presenting their computer science final projects on stage in front of an audience, with the hashtag "#csfinalproject" displayed prominently.
Computer science students at Frisco High School pitched environmental technology projects to industry leaders from American Airlines, Ericsson, and Goldman Sachs.
more work