Every paper we've published

A field-image dataset of 1,000 onion plants trained an upgraded YOLOv8 detector that spots Anthracnose, Stemphylium blight, Purple Blotch, and Twister disease at 77.3% accuracy and 123 frames per second — fast and light enough to run inside a farmer-facing smartphone app.

A survey of 400 Uttar Pradesh farmer households finds that regular users of digital advisory platforms — WhatsApp groups, YouTube, apps — score significantly higher on knowledge, empowerment, and yield than non-users, even after controlling for income and education.

A qualitative study of 37 New Zealand dairy and sheep/beef farmers finds they increasingly trust informal, digital, and peer-driven information — including AI — over formal institutions, and are struggling with real information overload as a result.

A multi-agent AI system that generates season-long farming advice and checks it against a crop growth simulator before recommending it — in a decade-long test on maize in Karnataka, all three reasoning strategies beat static advisory guidelines, with the best approach lifting yields by over 1,150 kg/ha.