Microsoft Unveils New AI Project Built for the Global Majority, Starting with Kenya:
Artificial Intelligence:
As Artificial Intelligence (AI) continues to amaze and intrigue people, especially those with sparse knowledge on & exposure to this digital innovation, Generative AI that powers apps and tools that boost productivity and knowledge in most places across the world continue getting readily embraced.
Yet, these systems do not work equally well for all communities, especially those under-represented online, where most AI training data originates, reports released on Tuesday November 18th, 2025, outlines.
As a result, generative AI performs poorly in many languages and does not reflect the social and cultural realities of every population. Infrastructure challenges are partly to blame, but in nations where low-resource languages dominate, adoption of AI is lower, even after adjustments are included for low penetration in GDP and internet access.
Commendably, Project Gecko comes in handy in these situations. Essentially, Project Gecko is a Microsoft Research-led initiative designed to close these equity gaps by creating cost-effective, flexible ‘tailorable’ AI systems that can and indeed deliver vital expertise to the global majority.
The initiative is powered by experts from Microsoft Research Africa in Nairobi, Microsoft Research India and the Microsoft Research Accelerator in the United States; Digital Green, and other partners across ‘agri-tech’, philanthropy, and academia. Together, the partners are developing cost-effective, tailorable or ‘customizable’ AI systems built for the global majority and which speaks local languages and also incorporates culturally relevant knowledge. The AI systems also engage users through text, speech, and video.
Central to this effort is the MultiModal Critical Thinking Agent (MMCT Agent), a new multimodal AI system that analyzes speech, images, and video to generate context-rich, locally grounded answers. The MMCTAgent is now available on Azure AI Foundry Labs, with open-source code accessible on GitHub.
This work aligns with Microsoft’s mission to empower every person and organizations to achieve more by advancing AI that is equitable, responsible, and representative of communities worldwide.
As Ashley Llorens, Corporate VP & MD of the Microsoft Research Accelerator programme, explains, “Building AI systems from the ground up shaped by the knowledge, languages, and modalities of the global majority yields more innovative, useful solutions for a greater number of people.”
Why Agriculture Comes First;
While Project Gecko is poised to expand into healthcare, education, and retail, the team chose agriculture as its starting point because the sector is a powerful economic catalyst. In Kenya, among other developing countries, and India, agriculture contributes significantly to GDP and employs millions, most of them smallholder farmers working less than five acres of land.
However, linguistic and cultural diversity presents major barriers. Farmers often switch between local languages such as English, Kiswahili, Kikuyu, Kalenjin, and various Indian dialects. Many rely on oral instruction and visual demonstrations, and in some regions, low bandwidth and limited device capabilities impact the accessibility of digital tools.
While existing apps attempt to offer agricultural guidance, researchers have found out that farmers frequently receive incomplete or inaccurate answers because most underlying models were trained primarily in English.
Tanuja Ganu, the Director of Research Engineering at Microsoft India observes, “Agriculture has very specific terms, which may change from language to language, and even district to district. All those domain-specific nuances need to be understood better.”
To address the above among other concerns, Project Gecko builds on Digital Green’s FarmerChat, a speech-first assistant used by extension workers to support millions of farmers. Remarkably, Digital Green has amassed over 10,000 agricultural videos in over 40 languages and dialects. Yet, historically, that knowledge was not easily searchable or accessible due to linguistic and technical limitations.
Rikin Gandhi, CEO of Digital Green, affirms that the unlocking this knowledge will support even more farmers to get real-time responses in their local language and preferred modality.
For instance in the Project, a farmer in Nyeri County, Kenya, East Africa, can now ask a question verbally in his local dialect (Kikuyu language) and receive instant answers in text, audio, and video in the same language. Such a farmer jumps directly to the exact time span & stamp where the solution appears.
MMCTAgent on Delivering Accurate, Community-Grounded Answers
Notably, the MMCTAgent enhances frontier models by allowing them to reason across audio, visuals, and text. It breaks complex questions into smaller components, verifies its own answers, and grounds responses in real-world agricultural practices captured in community-generated videos and transcripts.
Field studies in Kenya and India revealed significant improvements in accuracy, usability, and user trust compared to generic AI systems.
As Lakshmi Devi, a farmer from Bihar, India, shared: “Before this development, we would ask neighbors or dealers for advice and weren’t sure it was right. With FarmerChat, we ask our questions, follow the instructions, and see better results.”
Building New Speech Models for Local Languages;
Human-centered research by Microsoft in Kenya and India also showed that farmers strongly prefer voice interactions. However, many local languages lack the basic tools for automatic speech recognition (ASR), text-to-speech (TTS), and the translation, needed to power voice-enabled AI systems. Existing datasets were too limited to train robust models as well.
Project Gecko’s team has therefore developed these tools from scratch, training and fine-tuning them using local datasets. To ensure accessibility on low-cost devices common in rural areas, the team uses small language models (SLMs), (these are compact models that run efficiently on minimal computing power yet often outperform larger LLMs in specialized tasks).
The team has already expanded support to Swahili, and some Kenyan dialects such as Kikuyu, Kalenjin, Dholuo, Maa, and Somali through a dataset of 3,000 hours of crowd-sourced Kenyan speech and is creating a public leaderboard to benchmark African languages performance. Insights from more than 130 farmers are helping shape features such as clarifying questions, actionable recommendations, and tools that support peer-to-peer knowledge-sharing.
Looking Ahead;
Project Gecko underscores Microsoft’s commitment to building AI that is locally adaptable, culturally relevant, and equitable. But achieving population-scale impact will require rethinking how AI is localized, evaluated, and deployed in regions where connectivity and computing capacity remain limited.
By studying what works in agriculture, Microsoft aims to create design patterns and tools that can be replicated across healthcare, education, and other essential sectors. The team will soon release a multilingual playbook with practical guidance for developers creating domain-specific AI tools for the global majority.
As Ganu emphasizes, “Our goal is to ensure that the next generation of AI is not only powerful, but also globally inclusive, culturally relevant, and shaped by the communities it aims to serve.”
