Farming meets the future: This AI-powered app equips farmers for a changing climate

Digital Green, a global development organisation founded in 2008, leverages its AI-powered assistant, Farmer.Chat, to empower small-scale farmers by enhancing productivity and increasing their incomes. The result? A 10-fold reduction in costs, accompanied by a 24% rise in income levels. The company, which derives its name from a mix of technology and climate-resilient agricultural practices, aims to focus on capacity building and training farmers on a large scale to achieve the desired outcomes.

Partnering with governments and community organisations in various countries, the organisation uses the AI platform to boost the cost-effectiveness of public extension systems. In a freewheeling chat, Rikin Gandhi, CEO of Digital Green, and Nidhi Bhasin, CEO of Digital Green Trust, spoke to ET Digital about how the organisation is using new-age methods to win farmers’ trust and improve agricultural practices. Edited excerpts:

The Economic Times Digital (ET Digital): Tell us more about Digital Green and the genesis behind starting the firm.
Rikin Gandhi (RG): Digital Green started 19 years ago as a part of Microsoft Research in Bengaluru. At that time, even just outside Bengaluru, in the villages, BSNL was just installing their first cell phone towers. So, we took a very broad view of technology and started using videos by and for farmers to share relatable practices.

The videos featured fellow farmers demonstrating sustainable agriculture practices. For instance, instead of randomly sowing the seed where one ends up losing a lot of it and the productivity is low, one can try line sowing, and then the productivity per seed increases. Another example is how one can do practices for rice cultivation so that instead of flooding the field, a moist condition is maintained in the soil. So, less water is lost, and the result is more productivity. The videos also showed what fodder they can feed to the animals to help enhance protein content, which would ultimately increase the farmers’ milk yields.

Eventually, we spun off from Microsoft into an independent organisation and started partnering with multiple NGOs and government departments across India, and later, in various other parts of the world.

ET Digital: Before this app, what metrics did you use to assess the traction of this product? How did you determine the extent to which it was seeping in?
RG: The government extension agents who were showing the video on a mobile projector to a self-help group (SHG) of farmers once every fortnight would record who watched which type of videos to gauge the interest and data points. They also recorded the kind of questions that came in and the feedback. Later, they kept track of the farmers who adopted such practices on their farms.

This was how we scaled to about 6 million-odd farmers, observing a double adoption rate compared to the conventional system and a reduction in costs.

Especially during the Covid-19 pandemic, there was a lot of migration back to the villages, and we began publishing the videos on YouTube; we got about 90 million views of these productions, which were in local, amateur languages. We understood that we could take this existing library of videos, merge it with other types of data like weather and market information, and put it across as a knowledge base for farmers and extension agents to use directly on their phones. They could also use local languages, voices, texts and images to get answers to the issues they were facing in their fields.

iStock

ET Digital: You launched Farmer.Chat in 2023. Tell us more about how this launch panned out.
RG: Generative AI and mobile connectivity in these rural communities were expanding; many of these communities were already searching online. We saw almost 100 million people watching our videos on YouTube. So, we thought of creating a much better product experience that would enable them to access not just our videos but also weather, market, and other relevant information. Now, we are in the process of not just providing information but also making connections for these farmers with local farmer-producer organisations. This will enable them to avail seeds or be able to sell their produce more effectively.

Though this is a global endeavour, we are working specifically in India, Ethiopia, Nigeria, and Kenya. In these regions, we focus on not only the product and application but also partnerships with governments and other organisations on the ground.

ET Digital: What is the on-ground situation in other agriculture-centric economies, and how does it compare to that of India?
RG: To give some examples, Ethiopia’s productivity is lower than India’s, but 80% of their population is into agriculture. At the same time, their connectivity is much less than India’s. India has widespread mobile internet coverage, and it has one of the lowest data costs in the world, whereas Ethiopia has the lowest telecom penetration and one of the highest data costs. So, we have had to adapt our approach. We are not relying only on mobile applications, since everybody does not have a smartphone in Ethiopia. We are also working with a government programme there (Ethiopia) with the Ministry of Agriculture, which operates a farmer hotline accessible via a basic phone. We have the AI agent available to those call centre operators so that when people call in, they can talk directly to the AI system even without a smartphone application.

But there are similar challenges in all geographies; hence, there is a need for these AI systems to be tuned for the local languages and vernacular of the agricultural domain. For instance, in Bihar, the word ‘bigha’ means different units of area, depending on which part of the state you are from. That is also true in Ethiopia. They have, just like India, a huge diversity of different languages that the community speaks. So, we have to train these AI models using these local languages to understand the specific vocabulary related to agriculture. Thereafter, we have to connect with those specific programmes that the government has; like in Ethiopia, the government has programmes for helping youth to get access to land so that they can do income-generating activities like producing saplings that can be used for some of the rainforest rejuvenation programmes. So, yes, depending on what kind of context we are operating in, there is a lot of variation.

ET Digital: What is the mindset of farmers in India as well as globally regarding the adoption of technology? How would they be incentivised to use an app and offering of this nature?
RG: It is a good question. I think basically because of the great diversity of agriculture that is so location- and season-specific, people are looking not for generic advice, say, from ChatGPT or Wikipedia; they are looking for very specific advice produced by and for farmers in their vicinity which they might be able to relate with. Together with the power of generative AI, which can make those kinds of videos and other data sources, like weather and market information, way more customised and personalised for the individual farmer and farm context, it makes them excited about it.

Farmers can now choose what is most relevant to them, resulting in significant growth as they have access to on-demand resources at their fingertips. They do not need to wait for an expert to come to them; instead, they can just take a picture and, in their local language, ask, “I see some kind of insect on my crop. What should I do about it?” And they receive an answer that is not generic but is highly specific to their concerns.

ET Digital: Are there some sections within the app which are viewed the most by farmers? Can you give us a breakdown of that?
RG: The most popular aspect that people are looking for is pest and disease outbreaks, and how they use the app for that is by taking a photo. And often it is not just a photo but a voice note along with that, like the example I mentioned about the insect on their crop. Another one is about weather information. People are looking for both short-term forecasts to know if there is going to be rain that will be sustained or if there is going to be some extreme event like a flood or hailstorm. And then the last one is about the market. These farmers are not just growing this produce for the sake of producing it. Some part of their produce is going to be for home consumption, but they are also looking to sell their surplus at the market. So, they look for information like what is the nearby mandi price for the commodity and if they should hold their produce.

A lot of the Government of India’s focus recently has been on mobilising farmer-producer organisations since there are many smaller farmers for whom it is difficult to negotiate with the market. So, there can be such collectives of 500-1,000 farmers who can have a storage facility or shared transportation vehicle for pickups or distributing the seeds. Such programmes exist across the country, but there is a need to coordinate those interactions to understand who wants the seeds or the pickup via transport. Hence, we are integrating such features within the same Farmer.Chat application.

Getty Images

ET Digital: Tell us more about how the Farmer.Chat app can revolutionise the way farming is done in India. What distinguishes it from other agri-tech apps in the market?
RG: That is a great question, and one of the things that we have been working on with the Ministry of Agriculture is to say that there does not need to be one app that does everything. The reality is that there are a lot of exciting services and applications that the government has created. What we need are mechanisms for interoperability so that, for farmers, it can be a one-stop way to get information about relevant government schemes, pest or disease information, how to connect with their local farm producer organisation and the like. So, we have been working with them to co-create this network and also to see that AI can actually serve as a way to aggregate that information which is coming from different sources into one channel.

ET Digital: What lies ahead for Digital Green in terms of increasing farmer incomes, offerings, and expansion plans?
RG: We are looking to scale. Over the years, we have scaled to about 7 million-odd farmers; specifically for Farmer.Chat, we are looking to expand to another 5.5 million farmers over the next three years. We want to really see these farmers’ incomes increase not just incrementally but substantially so that farmers can decide to stay in agriculture as a major source of prosperity or if they want to diversify to other livelihood opportunities too. Another big thrust area for us is to create this whole technology and product stack in such a way that it is open source and a digital public good for other organisations to build on top of. The third piece is that information is only part of the puzzle for these farmers. They want to be able to translate information into action and then ultimately income that can increase their livelihoods and prosperity, so that is going to happen through these connections to the farmer producer organisations and other service providers.

ET Digital: From an Indian perspective, Nidhi, could you share a case study of a particular farmer who has benefitted from this model?
Nidhi Bhasin (NB): So, before we actually launched Farmer.Chat, we did some work on a chatbot, which was more on Telegram. So, in Rajasthan, there was an advisory which was pushed to say that the rainfalls are delayed. This information led farmers to sow their seeds about seven days later. Those who followed the advisory saw a 15% increase in their yield and a 20% reduction in water usage. This illustrates the effectiveness of the model.

ET Digital: What effect has there been on the increase in farmer income, especially in the Indian market?
NB: So, the numbers are very similar to global markets. On a scale, more than 50% of the number we can share originate from India. From an impact point of view, we are looking at about 60% women farmers using our application and advisories. So, basically, the numbers globally and in India are not very different.

Our focus has been on some of the states where a larger number of small and marginal landholders exist, Bihar, Jharkhand, and Odisha, to name a few. Now, we are also working a lot in Andhra Pradesh and Telangana, and as we are scaling up, we are looking at other geographies like Uttar Pradesh, Madhya Pradesh, Karnataka, and Maharashtra.

ET Digital: Is Digital Green also involved in training for farmers to understand the platform in a better way and use it to their advantage?
NB: Absolutely. So, it is not like we are just saying to download the app and do what you want to do. It is also about training them to ask the questions in the right way. What are the different features? How can they use the voice feature better? Or how can they use the photo feature better? But in terms of scale, our strategy is to work with a lot of frontline workers or other non-profit organisations or people on the ground so that they can also further get trained and train the farmers better.

This product has been tested and tried a lot in India. Right now, we have about 100,000 users in India, which is only in the state of Bihar. So, while this product is for other geographies too, we feel it will have the maximum usage and learnings from India.