By 2030, agriculture could contribute around $600 billion to India’s Gross Domestic (GDP), an increase of about 50 per cent over the contribution recorded in 2020. But to get there, we must unlock growth and productivity for the sector.
There are several challenges that limit the potential of smallholder farmers. These range from post-harvest losses due to inconsistent quality sorting to yield reduction caused by fragmented agronomic advice severely. Artificial Intelligence (AI) is proving to be a transformative solution for agriculture, as it helps with the shift from manual and reactive to precise, predictive and data-driven agriculture.
The recently-held Mint All About AI Tech4Good Awards, sponsored by Salesforce, recognised two companies for their work in this domain. Sickle Innovations Private secured the Gold award in the Best Use of AI in Agriculture & Food Security category for its AI-powered grading and sorting machines, while Dayatani Digital earned the Silver for its platform delivering hyperlocal, data-driven advice to Indonesian smallholders.
Watch the highlights from the Awards Summit below,
Gold Winner: Sickle Innovations
Sickle Innovations has a significant presence in India’s fruit and vegetable grading machine market. The company has helped improve manual quality assurance in the post-harvest phase. The existing grading systems were rule-based, and sorting systems (removing defective items) were entirely manual. Manual defect detection and labour shortages created significant hidden costs. For example, a single defective apple missed manually can significantly reduce the price fetched for the entire box at auction, leading to substantial financial losses for farmers and traders.
Leveraging the power of AI, Sickle Innovations could analyse thousands of images per second and make consistent sorting decisions based on surface defects. Unlike static rule-based automation, their AI model continuously learns from new datasets, adapting to diverse varieties of fruits and regional defects. For example, pomegranates from Rajasthan have a different set of surface defects than those from Maharashtra. This deep learning vision enabled the company to deliver higher accuracy, eliminate human error and ensure real-time quality assurance even in variable environmental conditions.
The integration of AI into their optical sorting systems yielded immediate and direct gains in efficiency and profitability. Earlier, manual sorting achieved only 60-80 per cent accuracy, often leading to financial losses. The process of sorting itself required a large labour force and often took hours per batch. After the implementation of the AI optical sorting system, accuracy improved to over 95 per cent. The sorting speed tripled and labour costs dropped by 40 per cent. Consistent grading improved market trust and post-harvest losses reduced by up to 20 per cent. This offered a boost to profitability across the entire supply chain.
Sickle Innovations’ success can be attributed to in-house production and software stack, which gives it control over design and execution. Having a large number of machines installed across the country provides access to diversified and unbiased datasets. This defect detection capability brings a shift by allowing systems to quantify defects in a given lot, thereby increasing transparency in the process. The company prioritises transparency and accountability, ensuring all AI decisions are auditable, with human supervisors available to validate results and maintain operational control.
Silver Winner: Dayatani Digital
The silver award winner in the category, Singapore-based Dayatani Digital, focused on a challenge being faced by smallholder farmers in Indonesia, who often lose up to 30-40 per cent of potential yields. This is caused by fragmented advice, poor diagnosis of crop stress, and the lack of localised, data-driven guidance, as agronomists are scarce and advisory services rarely account for micro-climate or soil variability. Dayatani identified this knowledge and access gap as a core constraint limiting productivity and income for over 33 million smallholders.
AI enabled Dayatani to deliver adaptive agronomy that human experts alone cannot provide. Their system uses machine-learning models trained on multi-source farm data including soil, weather, satellite imagery and even farmer chat logs to infer the drivers of yield variation. This allows the system to generate real-time hyperlocal recommendations. They use Large Language Models (LLMs) to contextualise these insights into farmer-friendly guidance in Bahasa Indonesia, making expert agronomy accessible at national scale.
The shift from generic to data-based advice has a significant field-level impact. Before, the advisory was generic and reactive, with less than 10 per cent of farmers following data-based decisions. After using the AI-based solution, the platform, known as Pak Dayat, provided field-specific nutrition and pest guidance, leading to an 18-25 per cent yield uplift and a 12 per cent reduction in input cost in pilot districts. Farmers’ query-to-solution time dropped dramatically from 48 hours to under 10 minutes, demonstrating a clear improvement in issue resolution speed. The system continuously learns from farmer feedback and remote-sensing data, redefining digital agronomy from ‘rule-based advice’ to ‘evidence-based reasoning’.
The recognition of Sickle Innovations and Dayatani Digital showcases the role of AI in moving agriculture beyond traditional methods.
