Chasing peak sugar: India’s sugar cane farmers use AI to predict weather, fight pests and optimize harvests 

By
A man standing in a field of tall green sugarcane looks down at his mobile phone. A tall metal structure rises above him against a bright blue sky.

Read this story in Hindi & Marathi

Nimbut, Maharashtra, INDIA – “Ready?” Suresh Jagtap asks, then turns and disappears into a field of tall sugar canes. 

Inside is a world of dappled green. The 65-year-old farmer bats rustling leaves aside, clearing a swift path to a thin metal structure rising above.  

His family has been farming in the area for generations – vegetables, fruit and, more recently, sugar cane. Over the years, climate change has made the weather more unpredictable and extreme, and raised the risk of pests and disease. 

Recently, Jagtap turned to AI for help, aided by scientists at the nearby Agricultural Development Trust (ADT) of Baramati and using Microsoft AI technology.  

The tall metal structure is a weather station. At the top are wind, rain, solar, temperature and humidity gauges. At the bottom, sensors in the soil measure moisture, pH and electrical conductivity as well as nutrients like potassium and nitrogen. The data is combined with satellite and drone imagery as well as historical data and analyzed to generate simple daily alerts via a mobile app: Water more. Spray fertilizer. Scout for pests. A satellite map pinpoints exactly where each action is needed. 

The goal is to do just the right thing at the right time to optimize growing conditions and claim the ultimate prize: a harvest when sucrose content in the cane is at its peak. 

Since planting this one-acre test plot on his four-acre farm six months ago, Jagtap and his family have been following the guidance religiously. Harvest won’t be till November 2025, but already they see the difference.  

“The growth is good,” Jagtap said. “The leaves are greener and the height is more uniform.” 

“Seeing is believing”

India is the world’s biggest producer of sugar cane but much of it comes from small farms like Jagtap’s in Maharashtra and Uttar Pradesh in the north. For small farmers, drought, floods, pests and disease can wipe out entire harvests and push farmers into debt and even to suicide. 

ADT Baramati was set up in the 1970s to help farmers in the drought-prone region adopt modern farming methods. Today, it has an active outreach field force and its researchers collaborate with institutions around the world. Its field agronomists have introduced local farmers to – among many other things – drip irrigation, which uses far less water than the traditional flooding of fields, soil-less farming, modern grafting methods and artificial insemination with foreign bulls and local cows to raise milk yields. 

Pratap Pawar, trustee at ADT Baramati, in a polyhouse, a kind of greenhouse that provides a controlled climate for plants. Photo by Selvaprakash Lakshmanan for Microsoft. 

Some 1.6 million local farmers are beneficiaries of ADT Baramati. The Trust hosts an annual farmers’ festival called Krushik at its 150-acre campus where new techniques are introduced, attended by more than 200,000 farmers from around India.

“Seeing is believing is the philosophy of farmers,” said Pratap Pawar, a trustee at ADT Baramati. 

It was at the January 2024 farmers’ festival that ADT Baramati unveiled its AI project – about a dozen crops from sugar cane to tomato to okra, all grown with insights harnessed by AI. They called it the “Farm of the Future.” 

The sugar cane test plot had yielded stalks that were taller and thicker – weighing 30 to 40 percent more at harvest – and yielding 20 percent more sucrose. The plot required less water and fertilizer, and the entire crop cycle was shorter – 12 instead of 18 months.   

“We showed water-related data, weather data, nutrients, pH of the soil,” said Dr. Yogesh Phatake, a microbiologist working on the project. “We got a very exciting response.” 

Some 20,000 farmers signed up. From those, 1,000 were chosen for the first trial, focused on sugar cane. An initial cohort of 200 began planting in mid-2024. 

The sugar cane test plot yielded stalks that were taller and thicker – weighing 30 to 40 percent more at harvest – with 20 percent more sucrose.

Averting an AI divide 

The technology brings in weather, soil and other data from satellites as well as farm sensors onto a Microsoft data platform called Azure Data Manager for Agriculture (previously called FarmBeats), so farmers can see precisely what’s happening at their farm with a few clicks. 

Project FarmVibes.ai, an open-source research project from Microsoft Research, builds on FarmBeats to analyze the data, along with historical crop data, to provide insights – covering everything from whether crops are getting enough water to whether a farm has a pest infestation, what kind and how to get rid of it. 

Generative AI takes it a step further. Microsoft Azure OpenAI Service turns technical details into simple daily actions for the farmer – fertilize in areas pinpointed by satellite data, for example, or scout for pests, all delivered through a mobile app in English, Hindi and the local Marathi languages. It also creates a crop lifecycle plan for farmers in simple language, so they know exactly what to do next, taking the guesswork out of farming. 

The mobile app is called Agripilot.ai, customized for ADT Baramati by Microsoft partner Click2Cloud.  

Farm of the Future: How it works

A man standing in a field of tall green sugarcane looks down at his mobile phone. A tall metal structure rises above him against a bright blue sky.
Weather stations are installed in local farms with wind, rain, solar, temperature and humidity gauges at the top. Photo by Selvaprakash Lakshmanan for Microsoft. 
Wires and a small device connected to the stem of a sugarcane plant and the soil underneath it
At the bottom of the weather stations, sensors measure moisture, pH and electrical conductivity as well as nutrients like potassium and nitrogen. Selvaprakash Lakshmanan for Microsoft.
A man in a laboratory working on some tests and experiments with flasks and equipment around him
Soil from each farm is also taken to a lab for testing, in this case conducted by microbiologist Yogesh Phatake. Photo by Selvaprakash Lakshmanan for Microsoft.
A split screen on the wall showing patches of green representing satellite imagery of the farms
The local farm data is combined with satellite and drone imagery as well as historical data to give an accurate diagnosis of the crop health across the farm. Photo by Trifilm for Microsoft.
Based on these insights, AI generates simple daily alerts specific to each farm via a mobile app. Agronomists at ADT Baramati review these alerts for accuracy before they go out. Photo by Selvaprakash Lakshmanan for Microsoft.
Just outside the village of Nimbut in Maharashtra, farmer Suresh Jagtap checks his Agripilot.ai app. A satellite map of his farm shows different shades of green denoting the health of vegetation. Photo by Selvaprakash Lakshmanan for Microsoft.
An old man inspecting leaves of a sugarcane plant closely
Suresh Jagtap acts on an alert sent to his mobile phone to scout for brown rust and to spray fungicide on the affected plants. Photo by Selvaprakash Lakshmanan for Microsoft.
A woman holding a phone and reading the instructions written on the screen
The mobile app has an AI chat function for asking questions. Farmers can also call agronomists from ADT Baramati directly. Here, farmer Seema Chavan asks question on weather conditions at her farm outside Baramati. Photo by Selvaprakash Lakshmanan for Microsoft.
A bullock cart loaded with sugarcane stalks after harvest
The ultimate goal: a harvest of sugarcane grown at peak sucrose. Photo by Selvaprakash Lakshmanan for Microsoft.

For some of the people involved, this one feels personal. 

Click2Cloud is based in the US and its other clients include giant US agribusinesses, Middle Eastern governments and Southeast Asian plantation companies. CEO Prashant Mishra grew up near Nagpur, a district in Maharashtra that has witnessed a high number of farmer suicides. 

“We are giving the small farmers the data, tools and intelligence which we give the big shots,” Mishra said. Besides Maharashtra, Click2Cloud is working to deploy the solution with state governments of Chhattisgarh and Uttar Pradesh for multiple crops. 

Ranveer Chandra, chief technology officer of Agri-Food at Microsoft headquarters who started the FarmBeats project back in 2015 and helped launch Baramati’s “Farm of the Future” in 2024, spent holidays as a boy at his grandparents’ farm in Bihar, northeastern India. 

“In the past, farmers have suffered from a digital divide,” Chandra said. “If we don’t do it right, we will have an AI divide. Rich farmers become richer, poor farmers get poorer.” 

Attaining peak sucrose  

In mid-2024, about 200 farmers around Baramati planted test plots of about an acre each, each paying a one-off soil-testing and training fee of 10,000 INR (US $117) to ADT Baramati to be part of the trial. The Trust kicked in 75,000 INR (US $882) in hardware and other costs per farmer.   

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.  

The Jagtaps noticed improvements early on. For example, each sapling produced 10 or more tillers– the shoots that develop into stalks– compared to five or six previously. 

Another local farmer, Seema Chavan, 54, planted sugarcane on one-and-a-half acres in August. Her remaining two-and-a-half acres is still farmed the conventional way. 

She said she used to fertilize everywhere whether it was needed or not, and water everywhere, to the point that it degraded soil quality. 

Now she checks on her farm from her mobile phone every day.  

Farmer Seema Chavan stands in a sugarcane field. Photo by Selvaprakash Lakshmanan for Microsoft.

Interviewed at ADT Baramati’s campus, she quickly tapped various icons on the Agripilot.ai app, pulling up satellite maps of her farm showing varying shades of green – to indicate if there is vegetation stress in any areas. That way, she can irrigate, fertilize or spray pesticides only in specific areas, which is better for the sugar cane and also for her purse. 

An alert on December 2 warned of risk of Brown Rust, a plant disease, and urged spraying with fungicide. A December 15 alert noted a lot of weeds and urged weeding. Chavan, who lives 15 miles from her farm, calls in the alerts to farm workers when she isn’t there in person. 

During the rainy season, she received alerts not to water the crops. Neighboring farmers advised her to water anyway. “I took the risk,” she said, and desisted. Her sugar canes 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, as well as historical data from other farms and from sugar processors. That peak period lasts just 20 days, said Mishra of Click2Cloud, after which sugar levels start falling. 

Historically, farmers have been paid by the weight of their sugar cane, but what sugar factories care more about is sucrose content. Said Mishra: “We help with the entire planning around that.” 

ADT Baramati CEO Nilesh Nalawade discusses the AI data from the sugarcane trial plots with Click2Cloud CEO Prashant Mishra. Photo by Trifilm for Microsoft.

Researchers say the app will get more accurate as more farmers use it and the AI is better able to analyze data patterns. For now, agronomists at ADT Baramati are reviewing the AI-generated alerts before they are sent out.  

In the last six months, between 10 and 20 percent of the recommendations were edited for accuracy, Phatake said. By the time this crop cycle is over in late 2025, he said, the system should be mature enough to require minimal human intervention. 

The next disruption 

If AI can help make farming in India more sustainable, maybe fewer young people will leave for the big cities.  

“The youth will be attracted to farming,” said Aditya Vilas Bhagat, 28, another local farmer who has planted a test sugar cane plot. “We are facing that problem.”  

Aditya Vilas Bhagat, who works on his family’s sugarcane farm, says they are using less water, fertilizer and pesticide by following daily alerts from Agripilot.ai. Photo by Selvaprakash Lakshmanan for Microsoft.

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 sugar cane in the nearby village of Korhale Bhudra. AI is just the latest technology the farm has adopted – after drip irrigation pipes, solar panels for power and drones for spraying fertilizer and pesticide. 

For proponents of AI in farming, this is just the start. 

“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.” 

Top image: Suresh Jagtap, a farmer in Nimbut, Maharashtra checks on the health of his sugarcane crop via a mobile app. Behind him is a weather station. Photo by Selvaprakash Lakshmanan for Microsoft.