New Delhi:
For centuries, unpredictability has been the bane of a farmer’s existence, with too little or too much rain, pests and disease all threatening to destroy crops, pushing him into debt and, in some cases, even driving him to suicide. With climate change, the unpredictability has only become more pronounced and made a farmer’s work that much harder.
Modern farming methods have helped alleviate the unpredictability to some extent and another technology, set to disrupt many fields, may hold the key to reducing it even further – Artificial Intelligence. Pilot projects have already begun in India and one such test is being run in Maharashtra’s Baramati tehsil, which is one of India’s biggest sugarcane producers.
The family of Suresh Jagtap, 65, has been farming in Baramati’s Nimbut village for generations, moving from producing vegetables and fruits to sugarcane, which brings in more money. But, to ensure the maximum return, it isn’t enough to ensure that the crop survives the vagaries of weather and pests; a farmer has to ensure lower input costs as well as harvest at the right time – a 20-day window – to get the highest-possible sucrose content.
Recently, Mr Jagtap turned to Artificial Intelligence for help, aided by scientists at the Agricultural Development Trust (ADT) of Baramati and using Microsoft AI technology.
On Mr Jagtap’s field is a weather station, a tall metal structure with wind, rain, solar, temperature and humidity gauges at the top and sensors at the bottom to measure moisture, pH (acidity or alkalinity), electrical conductivity and nutrients like potassium and hydrogen. The data from these is combined with satellite and drone imagery as well as historical data and then analysed to generate simple daily alerts sent to farmers via a mobile app – water more, spray fertiliser, check for pests. Exact locations are marked for the farmers on a satellite map of their farms.
While the harvest won’t be until October or November this year, Mr Jagtap says he has been following all the steps since planting a one-acre test plot on his four-acre farm six months ago, and can already see the difference. “The growth is good. The leaves are greener and the height is more uniform,” he said.
‘Farm Of The Future’
Pratap Pawar, a trustee at ADT Baramati, said the trust unveiled its AI project – called the ‘Farm of the Future’ – in January 2024 with approximately a dozen crops, including sugarcane and tomato.
“Seeing is believing is the philosophy of farmers,” said Mr Pawar.
The sugarcane test plot yielded stalks that were taller and thicker – weighing 30 to 40 per cent more at harvest and yielding 20 per cent more sucrose. The plot required less water and fertiliser, and the crop cycle was shorter – 12 months instead of the usual 18.
“We showed water-related data, weather data, nutrients, pH of the soil. We got a very exciting response,” said Dr Yogesh Phatake, a microbiologist working on the project.
About 20,000 farmers signed up and 1,000 were chosen for the first trial, focused on sugarcane. Nearly 200 of them began planting the crop in mid-2024.
Behind The Screens
On the software side, the technology driving the project brings in weather, soil and other data from satellites and farm sensors onto a Microsoft data platform called Azure Data Manager for Agriculture (previously called FarmBeats), so farmers can see what’s happening at their farm with a few clicks.
Project FarmVibes.ai, an open-source research project from Microsoft Research, analyses the data, along with historical crop data, to provide insights, including whether the crops are being watered enough and whether there is a pest infestation.
Generative AI (GenAI), in the form of the Microsoft Azure OpenAI Service, turns these details into simple actions for the farmer: fertilise in areas pinpointed by satellite data or look for pests – delivered through a mobile app in English, Hindi and Marathi. It also creates a crop lifecycle plan for farmers in simple language.
The mobile app is called Agripilot.ai, customised for ADT Baramati by Microsoft partner Click2Cloud, which is based in the United States and has clients like big US agribusinesses, Middle Eastern governments and Southeast Asian plantation companies. The company’s CEO, Prashant Mishra, grew up near Nagpur, a district that has seen a high number of farmer suicides.
“We are giving the small farmers the data, tools and intelligence which we give the big shots,” Mr Mishra said.
Ranveer Chandra, chief technology officer of Agri-Food at Microsoft headquarters who helped launch Baramati’s ‘Farm of the Future’, said, “In the past, farmers have suffered from a digital divide. If we don’t do it right, we will have an AI divide. Rich farmers become richer, poor farmers get poorer.”
Click2Cloud is also working to deploy the solution with state governments of Chhattisgarh and Uttar Pradesh for multiple crops.
Checking For Alerts
Mr Jagtap and nearly 200 other farmers around Baramati planted test plots of about an acre each, each paying a soil-testing and training fee of Rs 10,000 to ADT Baramati. The trust put in Rs 75,000 in hardware and other costs per farmer.
Mr Jagtap and his son, Tejas, 28, check their Agripilot.ai app every day for alerts. Tejas has a bachelor’s degree in agriculture and lives on the family farm with his wife. On the test plot, each sapling has produced 10 or more tillers – the shoots that develop into stalks – compared to five or six earlier.
Another local farmer, Seema Chavan, 54, planted sugarcane on one-and-a-half acres in August. She follows conventional farming techniques on the other two-and-a-half acres on her farm.
Ms Chavan said she used to fertilise and water all areas of her farm, whether it was needed or not, to the point that it degraded soil quality. Now she can also check on her farm from her phone.
Opening her Agripilot.ai app, she pulls up satellite maps of her farm showing varying shades of green – to indicate if there is stress in any areas. This lets her irrigate, fertilise or spray pesticides only in specific areas, which is better for the sugarcane, soil and her bottom line.
An alert on December 2 warned of risk of Brown Rust, a plant disease and, during the rainy season, she received alerts not to water the crop. Neighbouring farmers advised her to do so anyway. “I took the risk,” she said, and didn’t water the crop. Her sugarcane continued to thrive.
The final goal is to predict the best time to harvest, around October 2025. The Agripilot.ai app predicts peak sugar based on a combination of random testing on the local farm and historical data from other farms and sugar processors. That peak period lasts just 20 days, said Mr Mishra of Click2Cloud, after which sugar levels start falling.
Historically, farmers have been paid by the weight of their sugarcane, but sugar factories care more sucrose content. “We help with the entire planning around that,” said Mr Mishra.
For now, agronomists at ADT Baramati are reviewing the AI-generated alerts before they are sent out. In the past six months, 10-20 per cent of the recommendations were edited for accuracy, Mr Phatake of ADT said, adding that by the time this crop cycle is over in late 2025, the system should be mature enough to require minimal human intervention.
Preventing Migration?
Aditya Vikas Bhagat, 28, another local farmer who has planted a test sugar cane plot, said if AI can help make farming more sustainable, fewer younger people may leave to find work in the cities.
Bhagat has a Bachelor of Science in Agriculture and a Postgraduate Diploma in Agriculture Business Management, and returned to help oversee his family’s 160 acres of sugarcane in the village of Korhale Bhudra. AI is just the latest technology the farm has adopted – after drip irrigation, solar panels and drones for spraying fertiliser and pesticide.
“India is ready for this next disruption,” said Microsoft’s Chandra. “Think of the next wave beyond the Green Revolution. We are at the cusp of the next big disruption in agriculture with AI and data. India is early, but it is ready.”
