In Malawi, more than 180,000 rural households are supported by Ulangizi, an AI-powered chatbot that offers agricultural advice in Chichewa and English via WhatsApp. The tool, developed by Opportunity International with government backing, allows farmers to get answers about crops, pests and climate change adaptation. The system combines technology with local agents that facilitate access in communities without smartphones or stable connectivity.
The germ of a digital agricultural initiative
By: Gabriel E. Levy B.
A few years ago, in Malawi, agriculture was always the lifeline of the rural majority. Corn and cassava were grown, often under inherited techniques, without the certainty of yields when weather patterns began to mutate faster than the seasons.
Faced with this reality, Opportunity International emerged, an NGO that opted to use artificial intelligence as an accessible extension of knowledge.
Ulangizi started as a pilot between February and April 2024 with around 150 farmers, with support farmers called Farmer Support Agents, whose task was to mediate between AI and those who did not have a phone, access to data or sufficient literacy.
The tool operated via WhatsApp, in Chichewa and English, accepting text, voice messages or photographs to identify diseases, give advice on agricultural practices or suggest crops more suitable for the local climate.
Sustaining in changing terrain
The weather in Malawi was shaken by extreme events. Cyclone Freddy in 2023 left concentrated rains for about six months of storms in just a few days, with floods, damage to the land and the disappearance of crops.
To this were added dry episodes linked to El Niño.
This variability aggravated classic problems: eroded soil, lack of nutrients, pests, new diseases, lack of access to timely information.
In addition, many rural communities face structural barriers: low literacy, simple telephones or no reliable internet access, few government agricultural extension personnel.
In this scenario, late interventions cost enormous losses.
Each day of waiting could mean the difference between planting potatoes resistant to degraded soil or losing seeds.
Ulangizi was born to shorten those times and to adapt traditional knowledge and official guides to the time that farmers lived.
Ulangizi to the rescue of crops and economies
The most pregnant axis of this experiment is how AI, applied with human sensitivity, can reverse agricultural, economic and social devastation.
The case of Alex Maere serves as a stone of contrast. The 59-year-old farmer lost almost all of his corn crop after the cyclone, with only a handful of seeds left, about 8 kilos compared to the usual 850.
Thanks to Ulangizi’s advice, convinced by the local agent, he decided to plant potatoes, a crop that is more tolerant to the new affected soil.
With that harvest he generated more than 800 dollars in sales, enough to pay his children’s school fees.
That kind of economic return has multiple effects. It allows families to invest again, builds resilience to future shocks, and makes agriculture no longer just subsistence. In addition, the model of local agents, people who visit farms and help those who do not have a phone, allowed AI to be something not remote but close.
However, not everything is straight on the slope. Trust in AI responses is fragile. If you misidentify a plant disease, if responses are delayed due to signal problems, if the advice is not adjusted to the local soil type, it can lead to harm rather than help.
Echoes of other similar territories
Malawi is not unique in this quest. In India, the government has put in place AI-based systems to predict monsoon storms, sending SMS to millions of farmers in several states. This tool acts to anticipate variable rainfall, warn about the risk of floods or droughts and thus reduce predictable losses.
In other countries in Sub-Saharan Africa, similar initiatives are multiplying: applications that identify pests, prediction systems based on satellite data, networks of agro-extension workers who use basic telephones to disseminate knowledge. Private investments in agricultural technology also grew from about $10 million in 2014 to $600 million in 2022, reflecting that the market is beginning to see value in solving these climate and structural problems.
Tensions and central questions
The Ulangizi project illustrates a contemporary dilemma well: how to use artificial intelligence responsibly in extremely vulnerable communities. Because it is not enough for technology to exist, it must understand the linguistic, cultural, economic and agronomic realities. If not, it risks being unenforceable or even harmful.
One of the problems is infrastructure. Internet signal, electricity, phone charging, smartphone availability.
Rural areas often rely on intermediaries that bring devices to farmers, such as FSAs. That raises costs, puts logistical barriers, delays responses.
Another problem is the basic information: AI is fed by official manuals, databases, agricultural guides.
If these guidelines do not reflect local varieties, recent climate effects, or specific soil type, there may be generic responses that fail in particular terrain. Also the issue of model hallucinations: unexpected mistakes can lead to wrong advice, from wrong fertilizers to incorrect pest diagnoses.
And then there is the social aspect: the acceptance of the community, the trust in the tool. If a recommendation fails, it can destroy the credibility of not only AI but those who promote it.
Also, who pays for the data, who maintains the service, how to ensure the continuous updating of agricultural knowledge, climatic conditions, markets.
The financial and technical sustainability of Ulangizi when the initial funding ends or when it scales massively is a crucial question.
Light in the field: other concrete examples
Beyond Alex Maere, there are other stories that show what Ulangizi achieves when it works well:
Local support agents: Patrick Napanja, 33, is a farmer who became a support agent. He tours communities, carries a smartphone for those who don’t have it, teaches how to use the app or asks questions to the chatbot on behalf of those who need it. That human mediation has been essential.
Groups of farmers who meet weekly with local agents, in collective meetings, to share knowledge received, ask questions together and show what has worked and what has not. That not only amplifies usage but creates learning networks.
Local sales that improve after alternative harvests. Farmers who, after climate disasters, accepted suggestions for crops that were more tolerant to the new soil, such as potatoes or mixed crops, were able not only to cover consumption, but also to generate surpluses for sale. This allows them to pay school fees, reinvest in better seeds and acquire tools.
In conclusion, Ulangizi represents more than just a technological gadget. It is a bridge between official knowledge, local knowledge and climate urgency.
In Malawi, he demonstrated that AI, when designed with cultural humility, with supportive human agents and with real attention to infrastructure and trust, can offer a lever to recover crops, incomes and agrarian dignity.
But it also reveals that the challenges of precision, scalability and sustainability are not solved with good algorithms but with continuous commitment, with adaptations to the terrain and with the voice of the farmer at the centre. The future sows hope; Their germination will depend on how we take care of those seeds.
References
Opportunity International. “Agricultural Information Through AI.” Opportunity.org.
Microsoft. “Opportunity International empowers Malawian farmers with AI.” Microsoft.com.
Associated Press. “AI chatbot helps farmers in Malawi adapt to climate change.” AP News.
