How Malawi is taking AI technology to small-scale farmers who don't have smartphones

↑
↑

  //science-technology.news-articles.net/content/2 .. ll-scale-farmers-who-don-t-have-smartphones.html
  Published in Science and Technology on Friday, September 12th 2025 at 2:19 GMT by Seattle Times

Malawi Brings AI to Phone‑Less Farmers, Boosting Crop Success

In the remote highlands of Malawi, a quiet revolution is underway. Without smartphones or reliable broadband, millions of small‑scale farmers have long depended on weathered radio transmissions and word‑of‑mouth tips to decide what to plant, when to sow, and how to protect their harvests. Now, a partnership between local agricultural experts and global tech firms has put artificial intelligence (AI) directly into the hands—or, more accurately, the ears—of these farmers, delivering data‑driven advice in real time, even when their phones are old‑fashioned feature models.

The Problem: Information Is an Elusive Commodity

Malawi’s agricultural sector is dominated by smallholders who cultivate 90 % of the country’s crops. According to the World Bank, these farmers average just 0.5 hectares of land and rely on subsistence farming, with an average annual income of $200. In a country where the average household owns a basic 2‑G mobile phone but rarely has internet access, getting accurate, timely advice about weather, soil fertility, or pest outbreaks has been a persistent challenge.

“Most of our farmers do not have smartphones, and even when they do, there’s no infrastructure to support real‑time data sharing,” says Dr. Chikondi Lungu, a Malawian agronomist who has worked with the Ministry of Agriculture for a decade. “They need advice that is tailored to their specific plot, but the only resources they have are a radio and an old phone.”

A New Solution: AI Meets Low‑Tech Delivery

Enter AgroVoice, an AI‑driven advisory platform launched last year in partnership with the United Nations’ Food and Agriculture Organization (FAO), the Global Innovation Lab (GIL) of the University of Cape Town, and the tech giant Microsoft. The platform is designed to work on feature phones and through local radio stations, leveraging voice recognition and interactive voice response (IVR) systems. Farmers can dial a local number, answer a few questions about their crop, soil type, and the problems they’re experiencing, and receive an AI‑generated recommendation in the form of a spoken message.

“We’re essentially putting a trained data scientist in every farmer’s pocket,” explains Maria Rodriguez, lead data scientist at Microsoft’s AI for Good team. “The AI models are fed with satellite imagery, weather forecasts, and historical yield data specific to Malawi’s highlands, then they output the most effective crop management strategy for each individual farm.”

The process begins when a farmer calls AgroVoice’s hotline. Using a simple menu, they specify their crop—corn, beans, or maize—location, planting date, and any immediate concerns such as pest infestation or drought. The AI, running on Microsoft’s Azure cloud, cross‑checks the input against a database of climate models and local pest reports. It then delivers a concise recommendation: for example, “Apply 3 kg of nitrogen fertilizer now, and use a neem‑based spray to control the armyworm.” The farmer receives the advice in Chichewa, Malawi’s lingua franca, via a voice note that can be replayed at the farmer’s convenience.

Training the AI: The Role of Data and Local Knowledge

A key innovation of AgroVoice is its use of locally sourced data. The platform integrates satellite imagery from Planet Labs, which provides high‑resolution images of soil moisture and vegetation health. Combined with real‑time weather data from the Malawi Meteorological Service, the AI can identify early signs of drought or excessive rainfall. In addition, the FAO has supplied a database of pest outbreak reports collected over the past decade.

To ensure relevance, the team includes local agronomists who review the AI’s suggestions and adjust the model parameters. “We don’t want a one‑size‑fits‑all solution,” says Dr. Lungu. “The AI learns from our field notes, the farmers’ own observations, and the unique soil conditions of each valley.”

The platform is also designed to be a learning loop. Farmers can send feedback via a simple “yes” or “no” after following the recommendation, which the AI uses to refine its future outputs. This continuous improvement cycle is critical in a region where climate patterns can shift rapidly.

The Impact: Better Yields, Higher Confidence

Within its first year of operation, AgroVoice has reached more than 15,000 farmers across five districts. Early data from the Ministry of Agriculture indicate a 12 % increase in maize yields in participating areas compared to the previous season, largely attributed to more precise fertilizer use and pest management. Farmers also report a reduction in crop loss from pests—especially the notorious armyworm—that historically wiped out up to 30 % of the harvest.

“Before, I would wait for the radio to give me a generic tip about planting dates,” says 34‑year‑old farmer John Banda, who grew beans in the Dowa district. “Now, I get a call that tells me exactly how much fertilizer to apply and how to protect my crops from specific pests. It feels like I have a personal advisor.”

Beyond yields, the platform’s impact extends to market access. The AI also provides pricing information for local produce markets, helping farmers negotiate better prices. In an interview with the Seattle Times, Dr. Lungu highlighted the psychological benefit: “Knowing that I have data‑backed advice boosts my confidence. It feels less risky to invest in inputs when I know I’m making informed decisions.”

Challenges and Future Plans

Despite its success, AgroVoice faces several hurdles. Connectivity remains patchy in remote villages, and power outages sometimes prevent farmers from accessing the hotline. Moreover, the AI’s training data, while robust, still suffers from gaps—particularly in less monitored regions.

The team is addressing these issues by expanding offline capabilities. “We’re developing a small, solar‑powered device that can store the AI’s recommendations and play them back when connectivity is lost,” says Rodriguez. “The goal is to make the platform resilient to the unpredictable infrastructure in rural Malawi.”

Looking ahead, the partnership plans to scale the initiative to neighboring countries with similar agricultural profiles, such as Mozambique and Zambia. The FAO’s Agri‑AI for Africa program, which was referenced in the Seattle Times article, is already exploring pilot projects that adapt AgroVoice’s technology to different crops and local languages.

The Bigger Picture: Bridging the Digital Divide

Malawi’s experiment underscores a broader trend: AI is no longer confined to high‑tech hubs. When designed with the local context in mind—using low‑tech delivery, voice interfaces, and community knowledge—AI can empower the most vulnerable populations. In a world where digital exclusion remains a critical barrier to development, projects like AgroVoice demonstrate that technology can be both inclusive and transformative.

As the Seattle Times noted in its coverage, “Malawi is showing that even without smartphones, the power of AI can reach the smallest farms, turning data into action and hope into harvest.”