Farming Chatbot Helps Malawian Farmers Overcome Daily Challenges | Festina Lente - Your leading source of AI news | Turtles AI

In Malawi, an AI-based chatbot called UlangiziAI is transforming smallholder farmers’ access to important information for crop improvement and pest management. Available via WhatsApp, UlangiziAI enables farmers to get quick answers to agricultural questions in Chichewa or English, thus addressing barriers such as illiteracy and lack of reliable agricultural services in rural areas of the country.

Key points:

•  UlangiziAI offers personalized agricultural assistance via WhatsApp, using text, audio, and images.
•  The chatbot responds in English and Chichewa, languages accessible to Malawi’s rural population.
•  The AI technology is powered by platforms such as Gooey.AI and leverages Azure cloud and NVIDIA GPUs.
•  The app dramatically reduces farmers’ wait times for agricultural advice.

Malawian farmers with scarce resources recently began to benefit from advanced agricultural advice through the introduction of UlangiziAI, a multimodal chatbot that provides detailed and personalized recommendations on how to optimize agricultural yields and manage disease and pest control. This initiative, which has been active since February, allows farmers in Malawi to use their cell phones to interact with the chatbot through WhatsApp, making it possible to send questions via both text and voice notes and also including the option to upload photographs for visual diagnosis of crop-related problems. The system’s ability to understand and respond in Chichewa, the native language of about half of Malawi’s 20 million people, makes it particularly useful in a context where about 30 percent of the population is illiterate. 

Developed by the ONG Opportunity International, which has been working in Malawi for more than two decades, UlangiziAI not only provides answers, but offers help tailored to agricultural issues typical of the local context. Given the low prevalence of smartphones in some parts of the country, the NGO has employed a local support network of Malawians trained to directly support rural farming communities. These support agents receive smartphones and appropriate training to facilitate farmers’ access to the UlangiziAI app. 

Paul Essene, one of the developers of the technology behind UlangiziAI, points out how the chatbot effectively addresses the structural challenges that have always limited agricultural productivity in Malawi. The difficult living conditions of the majority of the population, who survive on less than $2 a day, combined with the country’s predominantly rural location, make access to specialized agricultural advice complicated. Government offices, which are supposed to provide technical support, are often difficult to reach and open irregularly. Many farmers have to walk long distances only to find that such offices are closed. In this scenario, UlangiziAI becomes an essential tool: thanks to the app, farmers can get timely answers on issues related to crops, fertilizers, fungicides, livestock, and agricultural techniques, thus reducing the waiting interval from days to seconds.

During the first three months of the trial, UlangiziAI received more than 4,000 questions in both English and Chichewa. Essene points out that the multimodal nature of the app allows it to effectively handle queries from different formats: text, voice, and visual. Thus, the app responds accurately and quickly, providing relevant information for farmers. The technology behind the chatbot is supported by Gooey.AI, a platform that combines AI-based workflows to improve interactions between farmers and the advisory system. The service is hosted in the Azure cloud and uses NVIDIA A100 Tensor Core GPUs to manage the data. One of the main challenges faced in the development of UlangiziAI was the lack of language resources in Chichewa, which required the implementation of specific techniques to optimize the chatbot. 

When a question is submitted via voice note, the system uses AI models such as Meta’s MMS Large or OpenAI’s Whisper 3 to transcribe and translate the message into English. If a farmer uploads a photo of a plant or pest, the image is processed by OpenAI’s GPT-4o model. The information collected is then combined and analyzed to generate a query for a vector database. The system queries a repository based on the Malawi Ministry of Agriculture’s Agricultural Production Guide, a 450-page document that contains data specific to the country’s agricultural context. Once the relevant data is identified, the chatbot translates the information and sends it to the farmer via text or audio on WhatsApp, including references and links to the original source.

Gooey.AI regularly updates the language and speech recognition models used by UlangiziAI to take advantage of advances in less documented languages. This approach ensures continuous improvement in the accuracy and effectiveness of the chatbot. So far, the app has been very well received by Malawian farmers. Looking ahead, Opportunity International plans to expand UlangiziAI to Kenya, making it available in local languages ​​such as Swahili.

UlangiziAI is providing Malawian farmers with a valuable tool to address the day-to-day challenges of the agricultural sector by improving the speed and accessibility of technical information.