This is an intervention statement by Rwanda’s ICT and Innovation Minister Paula Ingabire at the Africa Food Systems Forum in Kigali, August 31 – September 4, 2026. She was speaking during a high-level roundtable side event September 1, 2026. Read below:
A few sectors illustrate both the promise and the urgency of how we can use AI, and I believe agriculture is clearly an example that shows the promise of AI applications, but also the urgency to leverage AI as we go forward.
Let me start with a few statistics.
When you look at employment across Sub-Saharan Africa, nearly half of the employment we have across the region is in agriculture. And yet millions of farmers continue to make some of the most consequential economic decisions of their lives — what they plant, when they plant, how much fertilizer to apply, whether a crop is diseased, when to harvest and where to sell — with too little information, or often receiving information a little too late in the process.
For me, that statement of the problem is where I see the real opportunity of what AI can do for agriculture.
It is not about replacing a farmer. It is not about replacing agronomists. It is really about putting better intelligence in the hands of both farmers and agronomists.
Imagine, for a second, if a farmer could have access to an agronomist at any time — whether it is an agronomist, a weather expert or even a financial expert when they need funding.
Imagine if those experts were available whenever the farmer needed them, in a language that they understand.
How much potential and transformation would that provide to the farmer ecosystem, not just in Rwanda but across Africa?
In many parts of Africa, we are starting to see what the future could look like.
Just next door in Kenya, we have seen what PlantVillage Nuru does, where a farmer can use a smartphone to take a picture of a cassava leaf and identify whether the cassava plant has been affected by a particular disease.
CGIAR research found that the technology could diagnose cassava disease symptoms more accurately than farmers and agricultural extension agents.
Imagine what that means if we are able to scale such technology in places where there is not enough reach from agricultural extension workers across the farming ecosystem.
That example from Kenya shows the potential of AI to address the shortage of agricultural expertise while complementing what is already there.
In Zambia, they are tackling a different problem, where many smallholder farmers cannot access credit because they do not have a credit history.
What Apollo is doing is combining different kinds of data, from remote sensing and satellite imagery to machine-learning models. That information is then used to connect farmers to inputs, financing and the advice they need.
That really shows the potential of technology, and what AI can do in addressing financial exclusion for our farmers.
There are many examples that we could talk about. In many ways, they could sound like separate innovations because they are happening in different countries across Africa — from crop diagnosis to credit scoring, weather prediction, agricultural extension and market intelligence.
But in reality, they are not separate.
Together, they point towards a fundamentally different agricultural system in which every farmer can make better decisions, every extension officer can reach more farmers than they would otherwise be able to, financial institutions can better understand agricultural risks, and markets can become more transparent.
I know we have so many partners here, and I will be quoting some of the work many of you have been doing.
The World Bank has specifically identified around 60 potential AI applications across agri-food systems, covering areas including climate research, climate-resilient seed research, pest detection, precision agriculture, logistics and price forecasting, among others.
So our challenge today at this round table is not to demonstrate how AI can be used in agriculture.
The challenge is: What does scale look like?
We are seeing examples reaching hundreds and thousands of farmers. What will it take to take them to millions of farmers across Rwanda and across Africa?
Let me talk a little bit about Rwanda’s journey because we have learned a very important lesson along the way.
We did not start with AI.
In fact, we have only started talking seriously about AI, particularly in agriculture, over the last three years.
We started with the foundations that make AI useful.
Almost a decade ago, some of you in the room will recall when we launched the e-Soko platform, which is really about market prices.
This shifted us from manually collecting market prices from more than 30 markets across Rwanda to making sure those prices were easily available on a mobile phone that a farmer has.
Then came Smart Nkunganire, which is digitizing the agricultural input subsidy value chain.
Today, we have more than 1.5 million farmers registered on this platform, ordering subsidized seeds and fertilizers through a simple USSD transaction.
I will come back to why I emphasize the USSD interaction.
There are many other digital rails that the Ministry of Agriculture has invested in to make sure it can reach as many farmers as possible.
Each of these systems has addressed a specific problem.
But collectively, what we see is that they have created something much more valuable: a growing digital footprint of Rwanda’s digital agricultural economy.
Alongside this, we have invested in connectivity, digital identity systems for all and digital public infrastructure.
Where we are today is that we are looking at what AI can do because it will allow us to extract much greater value from these foundations that we have already put in place.
We will have a presentation right after these remarks, and we will look at some of the examples of what the Ministry of Agriculture is doing, such as Tunga, an AI-powered voice assistant being tested through the Ministry’s call centre.
Today, farmers can ask a question in Kinyarwanda.
Tunga draws from a validated knowledge base developed by the Ministry of Agriculture, the Rwanda Agriculture and Animal Resources Development Board and different partners.
During the test phase, what was clear and evident was that Tunga was able to answer 60% of the questions correctly. Where it is not clear, it escalates the question to a human expert.
Behind this Kinyarwanda speech benchmark is something fundamental: an AI model that cannot speak the language our farmers understand will never transform agriculture.
As we build these capabilities, we need to understand how to build solutions that adapt to our farmers, rather than forcing farmers to adapt to these systems.
And this matters for scale.
Rwanda wants to increase agricultural extension coverage from 35% in 2023 to 69% in 2029.
Three years from now, we want to be reaching more than 2.5 million farmers.
What that means is that we will continue to invest in expanding the number of agricultural extension officers, but hiring alone will not close that gap.
This is where technology becomes an enabling or multiplier factor, ensuring that we can reach more farmers with fewer extension officers.
With that background, I wanted to use some of my time during this opening segment to share with you how we are prioritizing AI for agriculture and the areas we are looking at.
The first — and we will get into the details during the presentation — is farmer advisory: making sure that personalized, timely and local-language advice is available to our farmers.
The second is crop and livestock intelligence, which means detecting pests, diseases and other risks much earlier.
The third is markets and finance: connecting farmers to buyers, making sure they get better prices, credit and insurance.
And the fourth is government intelligence, which will help us better understand production, anticipate shocks and direct input subsidies and other resources that could improve productivity.
The question for all of you as you listen to these priorities is: What is it going to take?
This is partly the reason why we have this round table, because this is where the partnership conversation becomes very important for us.
Across Africa, we have seen so many promising applications.
What we do not yet have is sufficient scale for the ecosystem that will support all these farmers.
As partners in this room, we need to understand what that supporting ecosystem should look like and how we can become, together, the ecosystem that farmers need.
I am going to leave you with about five areas where I feel partners in this room can rally behind to make this transformation different from what we have seen in the past.
First is investing in agriculture data foundations.
AI needs reliable local data from soil, weather, crops, livestock, markets and satellite imagery.
We need interoperable agricultural data infrastructure. Without that, we have siloed data that does not speak to one another, and people end up making the same investments to collect the same data.
That needs to be backed by appropriate data-governance frameworks and mechanisms that allow innovators to build responsibly on these datasets.
The second is investing in the African context.
When you look at the landscape, our farmers are growing crops that may be poorly represented in global datasets.
They also speak languages that are underrepresented in today’s AI systems.
Our farming systems, soils and climate conditions are different.
So we need African-language datasets, agricultural knowledge bases and locally validated models.
Some of these can be treated as digital public goods that innovators across the continent can build upon.
The third is financing the last mile.
A brilliant model that requires an expensive smartphone and continuous broadband will not solve the problem for many of our farmers.
Recall when I said earlier the importance of USSD interaction for our farmers and making sure we are building solutions that adapt to their context.
The solutions we build must work through voice, basic phones, extension agents and existing agricultural platforms.
Wherever possible, they should also work offline because not many of our farmers are connected as we would wish.
I cannot emphasize this point enough: when we build technology, we are building technology that adapts to the farmer, not demanding that the farmer adapt to the technology.
The fourth is working with us to move from pilots to scale.
This may be the most important request that I make today.
Africa has no shortage of pilots.
What we need is patient, multi-year financing to take solutions that are working for 10,000 farmers to millions of farmers.
That means financing integration with government systems, product development and improvement, computing capacity, farmer onboarding and local support, to mention but a few.
Each one of us in this room has a role to play in de-risking that transition.
And the last ask for me is investing in African capability.
We should not build agricultural systems that African governments and research institutions cannot maintain, evaluate and sustain over time.
Investments should therefore build capability alongside technology — through agronomists, data scientists, extension officers, farmers, researchers, policymakers and local technology companies.
That is how you create sustainability in the long run.
This round table is not another invitation to fund another set of disconnected pilots.
It is an invitation to build an AI-enabled agriculture system that is scalable, starting with Rwanda.
We have the beginnings of the digital infrastructure required.
We have defined the priority use cases that I mentioned.
We have government institutions that are prepared to integrate solutions into national agricultural programmes.
And we have a country that is small enough to move quickly, but with challenges that are representative enough that, if we can demonstrate success, Rwanda can become a proof-of-concept country allowing these solutions to scale beyond Rwanda.
Our proposition this afternoon to you is straightforward:
Pick a part of this system and build it with us.
Help us build the data foundation.
Help us build the intelligence layer.
Help us take solutions to our farmers because building solutions that our farmers are far from will really be a half-baked cake for us.
Help us rigorously measure whether these technologies increase productivity, resilience and outcomes.
And help us develop models that ultimately can travel beyond Rwanda.
That will be the proof point of the success of us convening here and doing this together.
I will end on this because the measure of success we are talking about will not be how sophisticated AI becomes.
It will be whether the farmer is able to plant at the right time, plant the right crop and have the right inputs required.
Whether she or he can detect a disease before losing a crop.
Whether they can get credit from a bank or financial institution that previously could not see them.
And whether they can get a fair price for their harvest.
These, I think, should be how we measure success when we use AI, so that it does not look like something disconnected from what productivity truly looks like.
This is the opportunity before us here as partners.
We are very grateful for you carving out time this afternoon to be with us.
We hope this will be the beginning of building something great together with you.
Thank you.
HERE IS VIDEO VERSION
