Our Published Research

Problem: Farmers often over-irrigate due to limited real-time soil data, wasting water and straining resources—especially under increasing climate pressure.

Solution: DASM uses advanced AI to generate high-fidelity soil models and filter noisy data, enabling smarter, risk-aware irrigation decisions that significantly reduce water usage without compromising yields.

Problem: AI models trained on generic internet data don’t understand how farmers actually think, speak, or make decisions—creating a critical "embodiment gap" that limits their usefulness in agriculture.

Solution: This research builds a novel framework that teaches language models to align with real-world, embodied farming knowledge—improving AI accuracy in tasks like tool use and soil assessment by over 60%, and enabling far more practical, farmer-ready AI tools.