Data Solutions for Tailoring Agronomic Support to Meet Regional Needs - with Tami Craig Schilling of Bayer Crop Science

Tami explains the universal plan-plant-grow-harvest cycle that farmers follow worldwide, regardless of crop or geography. She emphasizes how this consistent process creates opportunities for technology deployment across different growing seasons and regions, from South American multi-crop systems to US Midwest rotations and coastal orchards.

• •All farmers follow the same four-step cycle: plan, plant, grow, harvest—regardless of location or crop type
• •The cycle timing varies by region (e.g., two corn crops in South America vs. corn-soybean-wheat rotation in US Midwest)
• •Technology applications must adapt to this universal process while accounting for regional timing differences
• •Bayer's market development team bridges R&D and commercial sales to deliver products optimized for each stage

Discussion of extreme localization needs in agriculture, where soil history, farmer practices, and environmental conditions create unique requirements for every field. Tami uses personal farming examples to illustrate how historical land use (like buried structures) permanently alters soil profiles, requiring field-specific product recommendations.

• •Every field has unique soil profiles based on glacial history, previous land use, and management practices
• •Farmer practices (tillage vs. no-tillage) dramatically impact product requirements and management strategies
• •G×E (genetics by environment) interaction is critical—the same seed performs differently across locations
• •Pest pressure (weeds, insects, diseases) varies annually and geographically, requiring adaptive recommendations

Tami explains how generative AI enables Bayer to process complex matrices of genetic traits, environmental conditions, and pest pressures that exceed human cognitive capacity. The technology augments agronomist expertise by analyzing vast product performance data to deliver recommendations no single person could generate.

• •Traditional spreadsheets cannot handle the complexity of triangulating genetics, environment, and pest variables
• •GenAI augments human expertise rather than replacing it—combining data analysis with agronomist knowledge
• •The system must account for multiple seed genetics packages farmers use simultaneously, not single products
• •Bayer has used AI/ML in R&D for years; GenAI now extends this capability to farmer-facing recommendations

Exploration of how farmers provide critical local knowledge through prompts and questions, starting with ZIP codes but requiring detailed field-specific information. Tami emphasizes that farmer data remains their property, and Bayer's role is augmenting their decision-making with 24/7 access to insights beyond what sales reps can provide.

• •Sales relationships begin with farmers describing their operation, challenges, and what worked previously
• •ZIP code is the starting point, but farmers must provide additional detail for accurate recommendations
• •Farmer data ownership is respected—Bayer doesn't claim ownership of customer questions or field information
• •GenAI tools like ELI provide 24/7 access to insights when agronomists and sales reps aren't available

Introduction to Bayer's ELI (e-l-y) generative AI tool that starts with ZIP code localization and enables conversational engagement for product recommendations. Currently deployed internally to sales reps and being rolled out to R&D and market development teams globally, with plans for broader farmer access.

• •ELI uses ZIP code as initial localization layer, with optional opt-in/opt-out for privacy
• •Tool enables conversational back-and-forth engagement similar to human agronomist interactions
• •Currently launched internally to sales reps; expanding to plant breeding and seed production teams globally
• •Designed to augment human expertise by providing data-driven exploration capabilities

Critical discussion of how prompt quality determines AI output value in agriculture. Tami reveals that even agriculture experts need training on effective prompting, leading Bayer to develop prompt guides that help users specify pest stages, crop growth timing, and product requirements to get EPA-compliant recommendations.

• •How you ask the question fundamentally matters—prompt quality directly impacts recommendation accuracy
• •Agricultural jargon requires specific ontology work and model training to be understood correctly
• •Users must specify pest type, size/stage (2-inch vs. 10-inch weed, larvae vs. adult insect), and crop growth stage
• •All recommendations must stay within EPA label directions—prompting helps ensure regulatory compliance
• •Bayer provides prompt guides and regular training at national sales meetings to maximize tool value

Tami shares insights from Commodity Classic trade show conversations with progressive farmers who expressed uncertainty about autonomous equipment but showed interest in information-access tools. This reveals the importance of education and guides for technologies that operate differently than traditional farm equipment.

• •Even progressive, innovative farmers lack baseline understanding of how AI tools work
• •Autonomous equipment adoption faces knowledge barriers—farmers unsure whether to adopt without understanding
• •Information-access tools (like ELI) have clearer value propositions than autonomous machinery
• •Industry responsibility includes education, not just training—bringing farmers along with guides and regular communication
• •Once farmers understand how to provide details for personalization, they can't 'unsee' the value

Discussion of how human brains possess generational and contextual knowledge (like historical field use as pasture) that AI systems lack even with comprehensive algorithms. This highlights the ongoing need for human-AI collaboration and the limitations of pure data-driven approaches.

• •Human brains have context that platforms lack—even with complete algorithms and data
• •Generational knowledge (e.g., 'this field was pasture for years') impacts product selection but isn't in databases
• •Historical land use affects current pest control and product performance in ways data may not capture
• •This knowledge gap reinforces why AI augments rather than replaces agronomist expertise

Tami discusses how generative AI's scale neutrality enables support for both thousand-hectare commercial operations and 1-2 hectare smallholder farmers in developing regions. She emphasizes the economic development potential of providing advanced tools to subsistence farmers via mobile phones, even non-smartphones.

• •Technology is scale-neutral—works for 1-hectare smallholders and 10,000-hectare commercial operations
• •Smallholder farmers in Africa, India, and South America are customers deserving equal support
• •Mobile phone adoption (including non-smartphones) enables technology access in developing regions
• •Supporting smallholders drives economic development: stronger villages, education access, innovation opportunities
• •Weather unpredictability makes flexible AI tools especially valuable for resource-constrained farmers

Using current extreme rainfall in Southern Illinois as an example, Tami illustrates how unpredictable weather creates urgent needs for adaptive recommendations. Generative AI tools provide flexibility to adjust product choices and pest control strategies when conditions deviate dramatically from normal patterns.

• •Weather cannot be predicted—farmers need tools that adapt to abnormal conditions in real-time
• •Localized weather events (e.g., 20 miles away has different rainfall) require hyper-local recommendations
• •Extreme conditions change product needs and pest control timing dramatically
• •AI flexibility enables success even in unprecedented weather scenarios (like delayed planting)
• •This adaptability is critical as climate variability increases globally