The Center for Digital Agriculture celebrated the 2026 Champaign-Urbana AgTech Week by hosting two robust events centered around artificial intelligence in agriculture. First was the multi-day Precision & Digital Agriculture Hackathon, where students from multiple universities collaborated in teams of four to tackle real-world agricultural challenges through interdisciplinary problem-solving. Second was the CDA annual conference on Generative AI in Agriculture, which brought together experts and leaders from academia and industry to spotlight how cutting-edge AI technologies are transforming the future of farming.
“These events reflect what makes the Center for Digital Agriculture unique—bringing together students, researchers, and industry to translate advances in AI into real-world agricultural impact. From hands-on innovation in the hackathon to thought leadership at the conference, we are accelerating the transition from research to practice in digital agriculture,” said John Reid, CDA Executive Director and Research Professor at the Siebel School of Computing and Data Science and Department of Agricultural and Biological Engineering at the University of Illinois Urbana-Champaign.
Here are a few highlights to recap these events.
CDA Precision & Digital Agriculture Hackathon
Hosted at the John Deere Technology Innovation Center in Research Park on the University of Illinois Urbana-Champaign campus, March 6–8, 17 teams comprised of 75 undergraduate and graduate students spent 36 hours innovating solutions for four different tracks: Smart Crops, Smart Livestock, GenAI, and Analytics & Decision Support Track. The CDA Precision & Digital Agriculture Hackathon, sponsored by John Deere and ND AgTech Engine, provided participants with the opportunity to help shape the future of agriculture and address real-world challenges by developing solutions using foundational artificial intelligence, generative AI, machine learning, computer vision, and more.
“Our first Precision Digital Agriculture Hackathon brought together students from diverse backgrounds and training levels to address real problems in digital agriculture. The event was well attended, and the positive student feedback reflected the strong engagement and collaboration throughout the program. It was a promising start, and we look forward to building on that foundation in the next iteration of this hackathon,” said Isabella Condotta, Co-Organizer, Assistant Professor of Animal Science, University of Illinois Urbana-Champaign.
Below is an overview of the projects that won first place for each hackathon track.
🌾 Smart Crops Track
This track focused on applying digital tools to optimize crop production.
First Place Team: Illini Popcorn | Team Members: Ximin Piao, U of I PhD candidate; Prisha Singhania, U of I Grad Student; Priyal Maniar, U of I Grad Student; Yuyang Liu, U of I Grad Student
The Illini Popcorn team took first place with PlantFit. This irrigation watch system integrates a minimally invasive crop sap-flow sensor to monitor the “pulse of the crop”. It combines environmental data and forecasts to provide irrigation advice on when and how much to water the plant.
Expand this section to learn more about PlantFit – Irrigation Watch System
Drought is the leading cause of crop yield loss in the United States, accounting for nearly $1.9 billion in expected annual agricultural losses — nearly four times the impact of the next most costly hazard. Yet current stress detection methods remain fundamentally reactive: visual observation identifies damage only after photosynthesis has already ceased, soil moisture sensors measure the water supply rather than plant water status, and remote sensing provides periodic snapshots rather than continuous physiological monitoring. By the time stress is visible, the opportunity to prevent yield loss has passed.
PlantFit addresses this gap by shifting the sensing to the plant itself, using minimally invasive sap flow to monitor transpiration as a direct, real-time proxy for crop physiological status. Because transpiration rate is tightly coupled to stomatal conductance and photosynthesis, continuous sap flow measurement provides an early, quantitative signal of water stress before visible symptoms emerge. The system computes a Normalized Sap Flow Index (NSFI), sap flow normalized against concurrent atmospheric demand (solar radiation, temperature, relative humidity, and wind speed) — and establishes a field-specific baseline during an initial 7-day well-watered calibration phase. Stress alerts are triggered when the 4-hour rolling average NSFI drops below the 25th percentile of the time-matched baseline distribution, enabling detection of mild stress and timely intervention.
Sensor deployment is optimized for field-scale scalability using yield-zone-based planning and hypercube sampling, with wireless LoRa/Bluetooth connectivity and season-long battery life, eliminating installation overhead. The companion decision-support platform integrates sap flow data with local weather variables to diagnose probable stress drivers, including sustained high temperatures, precipitation deficit, and elevated wind speed and delivers actionable irrigation recommendations calibrated to the 48-hour precipitation forecast. Early stress detection has been shown to increase yield by up to 67% compared to visual inspection-based management. For a representative 500-acre farm, full system deployment is estimated at approximately $400 per acre, a cost recoverable within a single drought season given potential revenue losses of $400 or more per acre.
Future development directions include integration of satellite and multispectral imagery, retrieval-augmented generation (RAG) for deeper agronomic insights, nutrient uptake monitoring, and closed-loop automated irrigation control.
“Walking into my grad program, I never thought I would compete in an AgTech hackathon, explore a completely new domain, and take home the top spot. Stepping into this challenge allowed me to explore the vast amount of data in agriculture and the potential for using tech to drive tangible impact in this space. A huge thank you to the mentors whose guidance and feedback gave us the conceptual clarity to build this, and organizers for a phenomenal weekend of learning and networking,” said Priyal Maniar.
The Boilermakers Bushels team, comprised of Purdue University grad students Leonardo Bosche, Natalia Volpato, Pedro Cisdeli Magalhaes, and Gustavo Nocera, took second place with N Scout, a concept that combines expertise in agronomy, artificial intelligence, and data science for smarter nitrogen management in corn.
🐖 Smart Livestock Track
This track focused on enhancing animal welfare, health, and productivity with digital monitoring.
First Place Team: StatNERDS | Team members: Vanshika Namdev, U of I Grad Student; Allison Strackman, U of I Grad Student; Kritika Sukhramani, U of I Grad Student; Ting Yun Chen, U of I Grad Student
The StatNERDS team took first place with Farmer’s Friend, a conversational, AI-powered triage assistant that bridges the gap between symptom onset and professional veterinary intervention.
Expand this section to learn more about Farmer’s Friend
The critical shortage of food animal veterinarians, with only 3.9% of U.S. clinical veterinarians serving over 500 million food animals, leaves farmers without professional support during the vital 24–72 hour window of a health crisis. This diagnostic delay exacerbates animal welfare issues, increases the risk of infectious disease spread, and results in significant economic losses for producers.
Farmer’s Friend is a conversational, AI-powered triage assistant that bridges the gap between symptom onset and professional veterinary intervention. By utilizing a Retrieval-Augmented Generation (RAG) pipeline, the system grounds its assessments in agricultural extension and USDA health literature, ensuring that recommendations are safe and evidence-based. The platform features a user-friendly, zero-overhead interface that collects guided symptom data and optional photographic inputs to provide structured assessments. These assessments include ranked condition likelihoods, severity ratings, and actionable care steps. When high-risk conditions are identified, the system facilitates an immediate, pre-filled alert to local veterinarians to ensure rapid response. Furthermore, the app includes a persistent health dashboard that tracks individual animal histories, enabling longitudinal monitoring and more informed future decision-making. By empowering farmers with immediate triage tools, Farmer’s Friend reduces the risk of misdiagnosis, mitigates the spread of infectious disease, and promotes superior animal welfare outcomes without requiring advanced technical expertise, structured data, or constant proximity to a specialist.
“What I am most proud of is that we did not build a flashy chatbot. We built something designed to be fast, grounded, and responsible — with outputs backed by Illinois Extension and USDA veterinary documents, all in under 5 seconds. Huge thanks to the Center for Digital Agriculture, John Deere, and ND AgTech Engine for making this event possible,” said Kritika Sukhramani.
The Confusion Matrix team, comprised of University of Illinois graduate students Luana Benicio, Hasnat Md Abdullah, Mamunur Rahman, Cintia Araujo, and Jaqueline Braz, took second place with Reliable Livestock AI, an “AI-based feed bunk monitoring system” that combines image-based bunk scoring with uncertainty-aware decision support.
🖥️ GenAI Track
This track focused on harnessing generative AI to make agricultural data more accessible, interactive, and creative.
First Place Team: CatLovers.AI | Team Members: Zehua Yuan, U of I Grad student; Jingru Jia, U of I Grad student; Qianqian Du, U of I Grad student; Shujie Wu, U of I Grad student
The CatLovers.AI team took first place with AgOracle Precision Agriculture Assistant, a hybrid precision agriculture assistant that integrates LLM-based reasoning with locally calibrated statistical models to deliver actionable recommendations for farmers and agronomy advisors.
Expand this section to learn more about AgOracle Precision Agriculture Assistant
Farmers and agronomy advisors need fast, actionable answers to precision-ag questions, but there are two common gaps:
- General LLM answers can be helpful, but may not include local quantitative optimization.
- Statistical models can be accurate for specific scenarios but are too narrow to answer every question.
This project combines both: a general LLM path for broad agronomy guidance and a local statistical agent path for nitrogen-profit optimization when the query matches supported conditions.
Precision agriculture requires decision-support tools that can provide both agronomic guidance and locally optimized recommendations. However, existing systems often fall into two extremes: large language models (LLMs) that can answer a wide range of questions but lack quantitative optimization, and statistical models that produce accurate site-specific estimates but are limited to narrow scenarios. This project introduces AgOracle, a hybrid precision agriculture assistant that integrates LLM-based reasoning with locally calibrated statistical models to deliver actionable recommendations for farmers and agronomy advisors.
AgOracle uses a routed architecture in which user queries are first analyzed by an LLM to extract intent and contextual information such as location, year, and season. If the query corresponds to a supported nitrogen management scenario, the system triggers a local statistical agent that fits an ordinary least squares (OLS) yield response model and calculates the economically optimal nitrogen rate (EONR) using crop and fertilizer price information. When the query falls outside supported conditions, the system provides general agronomic guidance using the LLM.
The platform is implemented as a FastAPI backend with a React Native mobile interface, allowing users to interact with the assistant and test different decision modes. In a demonstration using on-farm experimental data from Central Illinois (Spring 2024), the statistical model estimated an economically optimal nitrogen rate of 185.7 lb/ac based on more than 1,300 observations. By combining AI-based reasoning with empirical agronomic modeling, AgOracle provides a scalable framework for intelligent decision-support systems in precision agriculture.
“Competing in this hackathon was an incredible journey for our team. We set out to prove that generative AI can do more than just chat: it can be practically grounded to deliver the highly precise, actionable solutions that farmers actually need after embedded with rigorous econometrics framework layer. We are deeply honored by the judges’ validation and thrilled by the real-world potential of our platform. We’re looking forward to developing this project further to make a tangible impact in the field, but first, we need to catch up on sleep and feed our very patient hackathon mascot, our cats,” said Qianquian Du.
The CropMIND AI team, comprised of University of Illinois students Sai Kiran Billa, Wendy Ma, Tracy Nguyen, and Mengelin Liu, took second place with cropMIND.ai, a multi-agent system that compresses the crop insurance pre-qualification process from days into seconds.
📈 Analytics & Decision Support Track
This track focused on turning diverse farm data into structured, actionable insights for producers.
First Place Team: Golden Tigers | Team Members: Cameron Jones, Tuskegee University Undergraduate Student; Kalon Jones, Tuskegee University Undergraduate Student; Meghan Franklin, Tuskegee University PhD student; Eniola Olakanmi, Tuskegee University PhD student
The Golden Tigers team took first place with CropVitals: A Composite Crop Health Intelligence System for Proactive Field Management.
Expand this section to learn more about CropVitals: A Composite Crop Health Intelligence System for Proactive Field Management.
Modern agricultural data exists in silos. Farmers and extension agents have access to satellite imagery, weather forecasts, and soil sensors — but no single tool integrates these streams into something actionable. By the time crop stress is visible in the field, yield loss is already locked in. CropVitals addresses this gap through an interactive field health monitoring dashboard that fuses three free, publicly available data sources — Sentinel-2 satellite imagery, NASA POWER atmospheric data, and USDA SCAN in-ground soil moisture — into a single Crop Vital Score. The system monitors seven crop vital signs, including NDVI, EVI, NDMI, and NDRE, normalizing each to a common 0–100 scale using min-max normalization and combining them into a weighted composite score. Index-specific thresholds grounded in peer-reviewed agronomic research from Auburn University’s Cooperative Extension System (ANR-3180, 2025) drive automated High, Medium, and Low categorizations and a three-tier alert level. Future development will expand the system’s capability through continuous full-resolution Sentinel-2 raster rendering, automated daily Google Earth Engine export pipelines, expanded SCAN station integration for spatial soil moisture variability, mobile-responsive field interfaces, and a machine learning pipeline supporting scenario planning driven by user inputs. The result is a shift from reactive to proactive crop management, enabling farmers and extension agents to detect stress days before visible symptoms appear, target interventions precisely, and make confident, data-driven decisions that reduce input costs and protect yield.
“Participating in the John Deere-sponsored hackathon at the University of Illinois Urbana-Champaign was a highlight of my experience. My team and I developed CropVitals, a crop health decision-support dashboard designed to provide farmers with accessible, data-driven insights using low-cost tools. Winning first place in the Analytics & Decision Support category was incredibly rewarding, and it validated the importance of building solutions that are both innovative and practical for real-world application,” said Cameron Jones.
The FieldNet team, comprised of University of Illinois undergraduate students Bobby Mandell, Griffin Burke, and Pranav Popuri, took second place with a precision agriculture prototype that optimizes grain harvest logistics through satellite imagery and discrete-event simulation.
CDA Conference on Generative AI in Agriculture
The energy was buzzing in the Chancellor’s ballroom at the Illinois Conference Center on Monday, March 9th, as over 170 people convened for the annual CDA Conference to discover how GenAI is redefining what’s possible in agriculture through insights from leading experts in industry and academia. Our featured speakers and presentations, including keynote talks by Brian Lutz, VP of Agricultural Solutions at Corteva Agriscience, and Stewart Collis, Senior Program Officer for Digital Agriculture Solutions at The Gates Foundation, explored how innovation, efficiency, and impact converge as we shape the next era of digital agriculture together.
The conference touched on various uses of AI and GenAI in agriculture, the collaborative efforts to ensure these tools and services are reliable and trustworthy, and predictions about the future of AI in agriculture. The agenda included perspectives from academia, farmers, and industry, including:
- Framing Research and Translational Opportunities in GenAI for Agriculture
- GeoSpatial Artificial Intelligence: Advances and Challenges
- GenAI Empowered Biological Molecule Discovery and Design
- Mining High-throughput Phenotyping Data with Self-supervised Machine Learning
- CropWizard + AI AgriBench
- Cultivating Spatial Intelligence: How Grounded Reasoning Empowers Autonomous Agriculture
- Ecosystems that Execute: Turning Agricultural Challenges into Field-Tested Solutions
- Farmer Perspectives Panel – Envisioning the AI-Equipped Ag Enterprise
- Information, Expertise, and Generative AI in Smallholder Agriculture: A Research Agenda
- GenAI Across Scales in Agriculture: From Molecules to Farm Decisions
- GenAI for Drone Imagery: The John Deere–Sentera Approach
- From Benchmarks to Fields: Evaluating AI for Smallholder Farmers
- Agronomy Intelligence (AI): Educate, Automate, Accelerate
- Future of GenAI in Production Agriculture Panel
WCIA Farm Broadcaster Stu Ellis attended the conference to learn more about the advances of AI in agriculture and sat down with keynote speaker Brian Lutz, VP of Agricultural Solutions at Corteva Agriscience, to discuss where farmers intersect with AI. Watch the interview here.
“AI is accelerating the future of agriculture, making this a defining moment for the industry. Corteva was proud to participate alongside peers, with UIUC convening industry and academia around the opportunities ahead,” said Lutz.
A crowd favorite and highlight of the conference was the Farmer Perspectives Panel, moderated by Tami Craig Schilling, Founder of DeepRoot Strategies and a member of the U of I Board of Trustees. The panel was comprised of Andrew Nelson, Farmer and Software Engineer of Nelson Farms, Kyle Courtney, CEO of Courtney Farms and RentEase and AI Smart Farms, and Heather Hampton Knodle, President of Knodle Ltd Family Farms. The panel discussed blending finance, investing, and a farmer-first approach to the value of data; building real-world bridges with tech leaders; and advancing precision agriculture while staying deeply grounded in community and farm realities.
“If AI is going to meaningfully impact agriculture, farmers must be part of shaping use cases, tools and data systems behind them. Their perspectives are essential to ensure these technologies solve real farm problems, operate transparently, and maintain farmer buy-in in how farm data is used and valued,” said Craig Schilling.
This event aimed not only to bring together leading experts from industry and academia but also to inspire the next generation of innovators, researchers, and scientists who will continue to address current and future real-world agricultural challenges.
“One of my biggest takeaways from the conference was seeing how interdisciplinary collaboration is actively shaping the future of agriculture. From AI-driven analytics to precision farming technologies, it was clear that innovation in this space is no longer siloed, but instead driven by the integration of computer science, engineering, and agricultural expertise. As a Computer Science student at Tuskegee University working in AI and agriculture, the conference reinforced my belief that technology can play a transformative role in addressing global agricultural challenges, especially around efficiency, sustainability, and food security,” said Cameron Jones, Undergraduate Student, Tuskegee University.
Thank you to CDA event sponsors, the National Center for Supercomputing Applications, John Deere Technology Innovation Center, and the NSF ND AgTech Engine for helping make these events possible!
Subscribe to the CDA mailing list to stay up to date on Center updates and offerings.