Agenda: AI Benchmarking in AgTech

Geospatial analytics in agriculture have mostly been built as one-off models - predicting yield, flagging a condition, answering a single question. In this lightning session, Tavant will share how we’re moving to general-purpose spatial AI trained on vast, aggregated multimodal and geospatial data for agricultural and livestock production. We’ll show how combining aerial and satellite imagery, public datasets, machine data, and points of interest can capture the real-world interaction between human and animal behavior, environmental conditions, climate, and local context over time - so spatial intelligence becomes reusable and is not rebuilt every time. You’ll leave with a clear view of how these learned representations enable high-fidelity, fine-grained spatial predictions with real-time inference, and where that matters most across agronomics, supply-chain optimization, policy planning, and disaster response.