AWS Startups Blog

Insights From Outer Space: Farming Smarter With OneSoil

OneSoil AI Smart Farming

For those who aren’t in the farming business, let’s start with one of the most common units in agriculture: the hectare. The area of a hectare equals 100 meters by 100 meters. That’s the same size as two typical European football fields side by side. With a purposeful stride, you can walk the perimeter of a hectare-sized field in about four minutes. You are not examining the crops, analyzing the soil, or taking inventory of irrigation equipment and machinery. You are simply tracing the edges of a hectare.

Now consider Lisbon, Portugal-based agronomist Francisco Castanheira. Castanheira is an olive tree specialist, or more precisely, an olive oil production specialist. In Lisbon, Castanheira gets a message from a massive olive oil producing farm in Azerbaijan. This isn’t one hectare, but thousands of hectares of olive trees. Like everyone he consults with, the Azerbaijani farmers want Castanheira’s help to increase the yield and quality of their product. In other words, they want Castanheira to analyze every tree and patch of soil so they can produce better olive oil.

A decade ago, Castanheira and other agronomists like him would have trudged or slowly driven through row after row of olive trees. That’s still the default for most farms around the world. Sounds nice, but not when you have thousands of hectares waiting for analysis from other clients.

In the last five years or so, rather than walking or driving it, you might pack in drones and fly the property capturing video and images that way. Better, but you still need a drone on site and likely someone to fly it.

Today, you could do what Castanheira actually did, analyze the entire farm using satellite imagery and some very smart machine learning models from Belarus-based startup OneSoil. And smart is the operative word.

The OneSoil service goes well beyond defining the boundaries of a farm or a field (which it does automatically in many regions around the globe to within a five-meter distance). With OneSoil (and some additional mapping data) Castanheira was able to examine the topography, the state of irrigation, the trees themselves, all in the context of historic weather conditions and typical yields in the region. Combining that data allowed him to identify a variety of problems in the olive groves and propose solutions. All without setting a foot in Azerbaijan. There would be plenty of time for trips to the South Caucasus later.

“I am able to have a clear picture of the property and the conditions before I get there,” Castanheira says while traveling from another client farm in southern Portugal to his home in Lisbon. “I can engineer my proposal and address the client’s needs far faster than I could have before.”

Clearly, it’s the combination of the data OneSoil provides and Castanheira’s own years of experience managing groves of olive oil producing trees that produces happy clients and better olive oil.

But the OneSoil team’s ability to get Castanheira in the weeds, as it were, from a data perspective via their cloud-based service, is what scales Castanheira’s brain. He can serve more people, and solve more problems, because OneSoil makes their data and their models easy to access no matter where a client is – and Castanheira’s clients could be scattered everywhere from Saudi Arabia to all across Europe. Client farms he consults with now can have olive groves 300 kilometers apart. “A single person, someone like myself, can manage very, very large areas,” he says.” Wherever they might be in the world.”

That is the mission of OneSoil, to bring what it calls “ground truth” to every patch of earth on the planet. Using satellite imagery as a starting point, the OneSoil team couples that with on the ground sensing of factors like soil conditions, temperature, and moisture levels to analyze and ultimately model the specific conditions that are ideal for growing whatever it is you might want to grow, or what you should be growing. Farmers use the service for free (though if they want to deploy sensors in their fields they have to pay a small amount for the hardware), and what they get in return is a detailed look at the state of the health of their fields.

For the farmers the insights OneSoil brings help them to answer the questions farmers have always had.  Given the inputs a farm has —  water, fertilizer, crop type, soil — coupled with changing weather conditions (especially given climate change), how can I best optimize my yield? How can I farm smarter?

In return for using the service, OneSoil uses the farmers’ data to train its neural networks to more accurately identify field boundaries and detect crops. With this data, governments are able to get accurate statistics about fields and crops within their borders. Others with a stake in the agricultural economy, whether it is buyers of crops or financial analysts, are able to better predict trends in local and global markets. Insurance companies can more accurately assess risks. And sellers of different farm commodities — seeds, fertilizers, pesticides and the like – can more easily discover new markets and new customers.

“For the farmers we can give them data as easily as sending us a message from their phones,” says OneSoil co-founder Slava Mazai from the startup’s HQ in Minsk. “We see several steps in the development of ‘agtech” through OneSoil. It starts with farm management system, analyzing the fields, understanding machinery, and all the inputs and tools in a farm. The second step is about insights, we want to deliver smart insights from the fields. So, if I am a farmer, I can get updated imagery from my fields every three to five days to see how my fields have changed because of something I did, or a condition that arose. I can take that image and consult with an agronomist.”

As the system gets smarter and smarter, Mazai and the OneSoil team imagine almost fully-automated farms, with underlying models tapped into irrigation, fertilizer, and harvesting machinery that will be able to predict and produce optimal yields; plan and monitor fieldwork; predict plant diseases and the emergence of pests: and generally, give recommendations to the farmer at all stages of work.

In the olive oil business, Castanheira sees a similar trajectory toward automation with the help of services like OneSoil. It doesn’t do away with the need for the expertise of agronomists like Castanheira, what it will do is free them to experiment more fully, to optimize and modify crops even further. “I believe that were it will go next is to try new varieties, new cross breeding, and all of it will be almost completely mechanized,” Castanheira says. “We are making more progress now in the olive oil industry than we have in the last 2000 years. And for farmers facing a climate that is changing so quickly and becoming so irregular, having more insight and more permanent crops like olives is an advantage to keeping the farming industry economically viable. Farming has always been hard, we need all the tools, data, and insight we can get.”