Multilingual EdTech · Nairobi, Kenya
AI-powered education in the languages Kenyan students actually speak. Umilisi AI makes the CBC curriculum accessible in Swahili and local languages for the first time.
"We are not asking for permission to build. We are asking for the infrastructure to scale what already works."
Umilisi AI was born from a fundamental truth: Kenya's Competency-Based Curriculum is delivered almost entirely in English yet for millions of students, English is not a first language. Language should be a bridge to knowledge, not a barrier to it.
We build AI tools that meet learners where they are. Our flagship product, CBC Msaidizi, lets students query the full CBC syllabus in Swahili for the first time. Our methodology is anchored in global research: UNESCO reports that in Africa, children learning in a familiar language achieve a 30% higher comprehension. Umilisi operationalizes this data for the Kenyan context.
And as one peer-reviewed study notes, "English is the prioritised language in education due to exams being set in English, the status of the language in local and global contexts, and the possibility of getting a prestigious job. Swahili is used to explain, elaborate, and enhance learners' comprehension of subject content. Swahili is considered more engaging for learners, as they understand the language better and can focus on the new concepts and terms of the subject matter."
At its core, Umilisi (Swahili for "mastery" or "competence") signifies more than just knowledge. It represents competence. It is the mastery that occurs when a student no longer has to struggle with translation and can instead focus on the essence of the subject. We build AI tools that ensure every Kenyan learner can achieve this level of Umilisi in their own tongue.
Generative & Agentic AI Category
Culture Preservation & Archiving Category · Jomo Kenyatta University of Agriculture and Technology
Top 15 out of 2,500+ submissions nationwide
A multilingual AI platform for Kenyan education and language access built from the ground up.
Featured Module
Kenya's Competency-Based Curriculum delivered in Swahili for the first time. Students across the country can now study their full CBC syllabus in their first language, without requiring English proficiency or a stable internet connection.
The problem
Over 10 million Kenyan learners interact with the CBC daily. The curriculum exists almost exclusively in English a second or third language for the majority of students.
The solution
CBC Msaidizi uses a fine-tuned Swahili-Gemma 4 model with RAG retrieval to answer CBC curriculum questions in Swahili accurately, contextually, in a personalised manner and at scale. Conversations are grounded in specific CBC chapters, with the model strictly scoped to curriculum content. All learning material is sourced from Opiq, a KICD-approved digital content platform.
"Language should never be the barrier between a child and their education."
Swahili RAG chatbot for Kenya's Competency-Based Curriculum. Students can query the CBC in their first language.
Swahili · RAG · CBCGeneral-purpose Swahili AI assistant. Everyday queries, explanations, and support in Swahili.
Swahili · General AISpeech transcription and translation across 19 Kenyan languages. From audio to text, no language left behind.
19 Languages · ASRA paginated digital library of Swahili literature content. Educational resources, searchable and accessible. Includes the first ever Swahili translations of children's classics like Treasure Island, Pinocchio, The Strange Case of Dr Jekyll and Mr Hyde, as well as more mature works like Leo Tolstoy's The Death of Ivan Ilyich, and The Duel by Anton Chekhov.
Swahili · Digital LibraryAchievements & recognition
Endorsement
Click to view full document
Endorsed by
Ubuntu Junior Academy
A school that integrates ICT into the CBC and has built a dedicated Talent and Technology Lab to advance digital literacy. Their endorsement comes from educators who work inside the curriculum daily.
Brian Githaiga is the founder and lead developer of Umilisi, based in Nairobi, Kenya. An AI engineer and Zetech University alumnus, Brian built Umilisi to address the language access gap at the heart of Kenyan education. Grounded at the intersection of NLP and classical humanities, Brian's work focuses on the formalization of Swahili as a primary medium for academic, linguistic and technical mastery.
His work has been recognised at the highest levels of Kenya's emerging AI ecosystem placing 3rd nationally in Generative & Agentic AI at the NIRU AI Hackathon Grand Finale, and winning the Culture Preservation & Archiving category at the JKUAT Interuniversity Hackathon 2025.
Brian leads all technical development of the platform, from model fine-tuning and retrieval architecture to interface design and deployment.
Founder & Lead Developer · Umilisi AI
Nairobi, Kenya
Nairobi, Kenya
"Bridging the gap between heritage and bleeding-edge tech, so every Kenyan learner can achieve Umilisi."