A green thumbs up for Agrolly

The 2020 Call for Code Global Challenge winning solution was developed for this reason. Short for “agricultural ally,” the Agrolly farming assistance app can help anyone who grows food take a more agile, data-driven approach to their work. Powered by open-source software on IBM technologies, the app provides short- and long-term weather forecasts, crop risk management insights, and an online environment for knowledge sharing and networking.

The team that built the app started out with one simple objective: to help others.

“We didn’t know exactly what to build and what challenges we would face. Call for Code is a competition with more than 400,000 participants, so we were scared,” says Manoela Morais, Cofounder and Chief Executive Officer of Agrolly. “We just knew that we wanted to help other people. That was our dream.”

Having met as students at Pace University in New York City, the four Agrolly cofounders came from different parts of the world. Morais was from Brazil; Ajinkya Datalkar, Chief Technology Officer, was from India; Chimegsaikhan [Chimka] Munkhbayar, Chief Operating Officer, was from Mongolia; and Helen Tsai, Chief Information Officer, was from Taiwan.

Inspired by the 2020 Call for Code competition’s theme of climate change, the team brainstormed a potential solution. Munkhbayar, who grew up helping her grandmother farm a small plot of land in the Sukhbaatar province of eastern Mongolia, talked about how farmers in the area struggled. The rest of the group shared similar scenarios from their home countries.

In India, for instance, agriculture is the primary source of livelihood for about 58% of the population, including Datalkar’s relatives. “In the Latur region in Maharashtra and other parts of India, farmers face droughts every few years, and the government is delivering water on trains. Many farmers end up committing suicide,” he says.

Known as the “Kingdom of Fruit,” Taiwan experienced its worst drought in more than half a century in 2020. In addition, seasonal rains are arriving earlier and later than usual, disrupting the flowering of trees.

In Brazil, small-scale farmers often can’t depend on farming insights handed down through their families quite like they used to, or even on last year’s personal experiences.

Learn more about the Call for Code Global Challenge.

The Agrolly team quickly found its calling, deciding to build an app for farmers in their and other countries. Then, the team members drew on their individual expertise to help make the project a success.

Munkhbayar, who earned an MBA in entrepreneurship at PACE on a Fulbright scholarship, took the helm on strategic business development. Using social media, she connected with several farmers and other agricultural experts in eastern Mongolia so that the team could gain a deeper understanding of their needs.

The team learned that the farmers had access to only short-term regional weather forecasts, and then only in document and PDF formats on scientific websites. The farmers needed yearly forecasts and multiyear trends specific to their farms. They also required easy-to-read weather data they could quickly download on their mobile devices, such as while in the field.

These and other insights helped guide the Agrolly team as it outlined the app’s core functions and technical requirements, including the ability to support older mobile phones, diverse languages and image uploads.

Next, the team got to work building the app. Datalkar, who had just completed graduate work in computer science and software development, designed the solution architecture and wrote the code.

“We were working 24x7. At one point, I was coding so much we made two major updates in one day,” he explains.

Morais, who is a chemical engineer with experience in agriculture risk management, handled the data analysis and project management. She helped the team adapt agile methodologies to fast-track their work.

“We had a list of priorities, and every day we met to assign tasks. Every one of us was responsible for one task, and if someone encountered a problem, we were all there to help him or her progress. Very tight planning helped make this project happen,” she says.

Datalkar agrees. “It was agile methodologies on an ultra-scale,” he says. “If I encountered a problem, such as with the design, I would just start calling the team and sending them screenshots, saying, ‘Take a look and tell me what you think.’”

Tsai has a background in business finance and is earning a graduate degree in computer science. She helped ensure the app presented information in a visually intuitive way so that users could quickly get the insights they needed to make educated decisions.

Working steadily, the team took advantage of IBM resources made available for Call for Code participants, including an IBM Cloud® account credit, IBM Watson® and other starter kits, a dedicated Call for Code Slack workspace monitored by mentors, and a technical content library.

“The IBM online product documentation is well done,” says Datalkar. “We saved a lot of time by using it. Also, mentors were always available on Slack to answer our questions.”

They developed and launched their MVP in 25 days. “It was intense, because Manuela, Chimka and I graduated on May 16. Then, on May 28 we all began working on the solution and had the MVP ready on June 26. It immediately went into the [Google Play app store] marketplace, where anyone in the world can download it for free,” says Datalkar.

They achieved their goal despite working during COVID-19 lockdowns and across vast distances. At the project’s start, all four team members were still in New York City, but shortly afterward Tsai traveled home to Taiwan and Munkhbayar visited Mongolia.

Flexibility was key, says Tsai. She recounts how sometimes she forgot about the time differences and texted the others when they normally would have been sleeping.

“It was two or three in the morning for them, and they would still respond and be supportive. I was like, ‘Oh, sorry, you're still awake?’ There was no sleeping the whole month,” she laughs.

The Agrolly app can deliver the insights smallholder farmers need to effectively plan crop production and minimize risks caused by unpredictable climate conditions.

First, the weather module provides a range of forecasts for the user’s village and province. The app displays hourly and weekly forecasts by connecting to technologies from The Weather Company®. Finding an affordable way to deliver long-term weather data proved more challenging. The team talked to agronomists and data scientists and extensively researched relevant data sets. They ultimately decided to write their own algorithm and pull five-year historical data from the NASA Global Precipitation Measurement database to predict long-term forecasts.

The second module enables smallholder farmers to perform automated risk assessments, representing a first-time industry innovation. It combines Agrolly’s long-term weather forecasts with crop water requirements published by the Food and Agriculture Organization (FAO) of the United Nations, which are based on location, type of crop and stage of growth. Using the Agrolly crop-risk model, it then delivers insights to help farmers determine what to plant and when to plant it. Farmers can also use the results to help qualify for loans and credit from local financial institutions.

“When a farmer accesses our app and selects a crop type, the planting date will begin and the location. The app then calculates all the location-specific risks related to climate, such as temperature and water needs, for that specific crop over the plantation period,” says Morais. “With this information, the farmer can make better crop decisions, such as selecting the best crop type according to the risk factors, or plan ahead for the necessary resources to deal with these risks — and with that obtain better productivity.”

The third module is an online forum where farmers can easily interact with others to learn what they are planting and share agricultural tips, either by texting or uploading images. This capability is highly valued in places such as Mongolia, where young household farmers starting out can learn from experienced elders. The team also anticipates launching an Ask an Expert module, which farmers can use to directly consult with agricultural extension specialists and access materials on best practices.

Initially, the team didn’t have access to server or cloud infrastructure to run and store statistical regressions for its algorithms. After further research, Datalkar found that the team could build the solution using the IBM Cloud Kubernetes Service platform and run regressions with the IBM Watson Studio model automation solution. It could then store the results on the IBM Cloud Object Storage platform before caching them on the Agrolly server. The Agrolly app also employs the IBM watsonx™ Assistant solution, which help users locate app tools and manage their to-do lists. The team will also use it to build the Ask an Expert function.

The team’s hard work paid off. Announced in October, the 2020 Call for Code Global Challenge grand prize (link resides outside of ibm.com) went to the Agrolly solution. As the winner, the Agrolly team received a USD 200,000 grant; tools to test and build out the platform; and ongoing support from IBM Service Corps, other technical experts and the Linux® Foundation.

Since then, the Agrolly team has formed its own business and is blossoming. “Winning the competition was a major transformation for Agrolly, because it gave us a real chance to form a business,” explains Morais. “We did it with help from IBM, who has provided many people to coach and support us. Also, publicity around the prize is helping make us more relatable in the industry.”

Eventually, in collaboration with the Linux Foundation, Agrolly will enable people around the world to help improve the app. “I’m building an interface and functions through which developers and others can make contributions,” says Tsai. “Even farmers will be able to share things like their crop knowledge and what challenges they face.”

The team learned from early user feedback that agricultural advancements, key crops and weather conditions vary across geographies, requiring the global app to be localized for Mongolia, Brazil and India. Tsai is also taking the lead on working with IBM to build a web-based interface for use in Taiwan. Expected to launch by early 2022, the new interface will enable Agrolly to test the effectiveness of web-based over mobile solution delivery.

Having tested localized apps in Mongolia, Brazil and India, the team is launching each country’s respective app, which can be downloaded for free on app marketplaces. Despite growing use of smartphones in emerging Asian countries, the Agrolly solution is the first farming app available in Mongolia. To encourage app adoption by local farmers in the eastern provinces, Munkhbayar works with regional agricultural specialists to provide group and in-house training sessions, conduct field research and respond to user feedback.

In Brazil, Morais is establishing relationships with universities and government agencies willing to help roll out the app in their regions. These organizations can more easily identify and address barriers to adoption among farmers in their communities, including poor digital literacy skills and internet access.

Furthermore, the Agrolly team continually tests and improves the long-term weather forecasting model. Datalkar is also moving the solution to a new architecture built entirely on IBM Cloud, which will allow Agrolly to affordably scale the app based on usage.

Other planned enhancements will allow financial institutions to calculate farmers’ credit scores and facilitate loan approvals. Farmers will also be able to use the app to move paper-based records to the app, such as for crop production and sales and for pest management.

Farmers using the app can currently run crop-risk analyses by location and growth stage for more than 50 crops, and Agrolly continues to add information on more crops. More than 1,600 people have downloaded the apps to date, including 500 people using the global app, 1,000 people using the app localized for Mongolia and 100 people using the app localized for India.

The farmers report that with longer-term weather forecasts and reliable risk assessments, they can make more informed decisions about field irrigation and other aspects of resource utilization. They can also more easily collaborate with banks to secure loans.

In addition, farmers appreciate that they can consult with agronomists and study new crops and agricultural practices, including by viewing photographs, while in their fields. Some farmers also anticipate using the app’s forum to mobilize into agricultural cooperatives, which can compete more effectively in marketplaces.

While locally relevant, the app can have larger impacts, too. Productive yields enable farmers to not only better care for their families but also contribute to national economic growth and food security. At a global level, helping small-scale farmers deal with climate change is crucial to driving several UN Sustainable Development Goals, including climate action, no poverty and zero hunger.

“We want to make agriculture more sustainable, so farmers can use their country’s water, soil and other natural resources more effectively,” explains Munkhbayar.

Honored to win the competition, the team continues to work hard toward helping farmers take greater control of their own destinies. “What I appreciate most is farmers’ gratitude, when you come through with valuable insights on how to help improve their future and proactively reduce their risks,” says Morais.

Based in New York City, Agrolly (link resides outside of ibm.com) provides a farming assistance app that offers farmers worldwide insight into weather, water and soil conditions required for productive crop yields. Agrolly aims to fill in the information gap so that farmers with fewer resources available to them can make more educated decisions, obtain the necessary financing and improve their economic outcome.