The internet has long been our go-to tool for finding information. From academic research to the nearest nyama choma joint, we have relied on keyword-based search engines like Google. However, this traditional model is being fundamentally reshaped by the rise of large language models (LLMs) such as ChatGPT, Claude, Gemini, and Mistral. These models are changing how we search, how we consume information, and ultimately how we make decisions.
As a technology expert in Kenya, I have seen firsthand how digital adoption is growing, especially among the youth and small businesses. But with this growth comes the demand for more precise, context-aware, and faster information access. That is where LLMs come in.
From Keyword Queries to Natural Conversations
The first major shift is how users frame their searches. Instead of typing disjointed keywords like “best phone under 20K Kenya 2025,” users now ask natural questions like, “What is the best smartphone for under 20,000 shillings in Kenya this year?”
LLMs are built to understand context, nuance, and user intent far better than traditional search engines. This eliminates the need to scroll through multiple links or filter through sponsored content. Instead, the model provides a summarised, actionable answer in seconds.
For example, a Kenyan student looking for the latest HELB application deadlines or government-sponsored university fees structure no longer needs to navigate poorly updated institutional websites. With an LLM like ChatGPT, they can simply ask the question and get an accurate answer drawn from multiple up-to-date sources (if connected to the web).
Transforming Local Business Discovery
Search engines have long prioritised global content creators, often sidelining local businesses that may not have optimised their websites for SEO. LLMs are changing this by surfacing hyperlocal information based on user input, not ranking algorithms.
A boda boda rider in Eldoret can ask, “Where can I buy a reliable second-hand smartphone in Eldoret under 10K?” and, depending on the model’s web capabilities, get aggregated data that includes informal marketplaces like Jiji Kenya, Facebook Marketplace, or even TikTok shops.
Similarly, a farmer in Meru may ask, “Which fertiliser is best for tomatoes in cold weather?” and get a targeted, research-backed response, instead of generic articles on agriculture from unrelated regions.
Redefining Learning and Professional Research
Kenyan professionals, especially in fast-moving industries like fintech and healthcare, rely heavily on updated information. Traditional search may present ten different links with varying credibility, forcing users to sift through content.
In contrast, LLMs can summarise complex information, reference source materials, and even compare options — all within one interface. For instance, a software developer trying to understand the best approach for mobile money integration with M-Pesa can now get code examples, API documentation summaries, and sandbox test recommendations without switching tabs.
Students preparing for KCSE or university exams are also using AI tutors powered by LLMs. These tools can generate quizzes, explain past paper questions, and offer revision tips tailored to the Kenyan curriculum.
Risks and the Need for Local Adaptation
While the benefits are substantial, there are risks. Most LLMs are trained predominantly on Western data. This means they may misrepresent Kenyan-specific contexts or fail to retrieve locally relevant sources unless fine-tuned or linked with local content repositories.
For example, an LLM might recommend tax advice based on IRS regulations in the US rather than the Kenya Revenue Authority’s guidelines. Or it may misunderstand Swahili nuances or misinterpret cultural practices.
This is why localisation and policy intervention are vital. Kenya’s growing innovation hubs, like iHub and Gearbox, should lead efforts in training and fine-tuning models on local data. The Kenya ICT Authority, Communications Authority, and Ministry of ICT should also engage with global AI providers to ensure equitable representation and data sovereignty.
Integrating LLMs into Kenyan Digital Life
Looking ahead, the integration of LLMs into mobile applications, government services, and education platforms can massively increase productivity and access to information. Picture Huduma services with a smart AI assistant guiding users through application processes. Or a KRA chatbot that not only answers tax queries but also helps file returns in Sheng.
We must, however, invest in AI literacy, data protection frameworks, and local model development to ensure these tools empower rather than marginalise.
Large language models are not just enhancing search; they are reimagining it. In Kenya, where time, data costs, and clarity matter, the adoption of these tools offers a chance to bridge the digital divide and unlock a new era of informed decision-making.
Whether you are a university student in Kisumu, a small trader in Gikomba, or a nurse in Turkana, LLMs are poised to become your next search engine — only smarter, faster, and more personalise
