Shaping the Deep Tech Revolution in Agriculture 2025

Data collection in smallholder farming ecosystems is a critical barrier to high-quality service delivery. In many economies, agricultural data collection is still manual and paper-based or conducted using basic software, and is prone to errors and inefficiencies. These systems also result in the exclusion of farmers with low literacy. To address this, the Wadhwani Institute for AI developed AgriAI Collect. AgriAI Collect uses automatic speech recognition to transcribe multilingual voice inputs, LLMs to extract structured responses and a human-in-the-loop system to validate low-confidence entries. Data is time- stamped and securely stored via a cloud-based platform. The Institute has onboarded 32,000 users; the data collected will be piloted for use cases such as organic certification. Such applications show the potential of GenAI to improve agricultural data collection, which can further enhance agricultural service delivery. CASE STUDY 1 Improving agricultural data collection and quality through GenAI – Wadhwani Institute for Artificial Intelligence, India Computer vision is a subset of AI that enables image-capturing systems to derive information from visual inputs including videos and images. By converging with machine learning, such systems can provide recommendations based on visual cues, reducing the need for human analysis. In agriculture, computer vision can be used to detect plant diseases, weeds or pests and to monitor crop stress in real time. Computer vision is also the foundation of autonomous systems including robots or agricultural grading/sorting systems.3.2 Computer vision Shaping the Deep-Tech Revolution in Agriculture 12