EqualyzAI, an artificial intelligence startup that wants to democratise AI innovations to help native African speakers access services in their local languages, has officially launched its solutions. The startup says it has a bold mission: to build AI solutions that solve real-world problems in emerging markets.
The company hopes to achieve this by using hyperlocal multimodal datasets, created in collaboration with native language speakers, to build localized models for agentic AI innovations. Thus, it aims to make AI truly understand African contexts and speak their languages, ensuring AI innovations are accessible, inclusive, and impactful for the communities that need them most.
The solutions launched by EqualyzAI are designed to address critical challenges in Africa while promoting linguistic diversity. They include:
uLearn: An LLM-powered tool for teachers to generate lesson plans, notes, and quizzes in English and local languages, revolutionizing education delivery across Africa.
DataCollect: A platform designed to collect hyperlocal datasets needed to build AI solutions that speak many African languages, ensuring inclusivity and representation.
AfroSLM 1.0: A Small Language Model (SLM) optimized to improve financial literacy in Yoruba and Hausa, empowering users with accessible financial education in their native languages.
Speaking during the launch on the International Day of Mother Languages, Co-founder and CEO, Dr. Olubayo Adekanmbi said: “Our mission is to ensure that AI speaks our languages, understands our contexts in text, speeches, images and videos, and solves our unique challenges. This is not just about technology; it’s about empowerment and inclusion. At EqualyzAI, we are bridging this gap by creating AI solutions trained on African realities, ensuring that no one is left behind.”
EqualyzAI wants to make AI understand Africans
Founded in January 2023 by Dr Olubayo Adekanmbi (CEO) and Dr Ife Adebara (CTO) AnalyzAI is an AI startup focused on building artificial intelligence solutions and generating AI possibilities for native language speakers across Africa. This revolves around leveraging hyperlocal data which means data that reflects the way the everyday speakers of these languages use the language in everyday conversations.
Using such data, the startup has built powerful small language models that can be integrated into products and services to reach last-mile users. The startup is looking to process over 75 African languages each spoken by at least a million people.
Explaining the purpose behind the establishment of the startup, Product Development Lead at EqualyzAI, Uche Edwin Okay, said the team discovered that existing chat models and chatbots were not performing as well as they should with African indigenous languages.
See also: Artificial intelligence has become a general-purpose technology- Olubayo Adekanbi
Indeed the team demonstrated how both ChatGPT and Google Translate failed to translate or provide the right and accurate translations that demonstrate accuracy and contextual understanding of local African languages, from Yoruba, Hausa, Igbo, and by extension some of the other major languages across Africa.
“And so we saw that that was a problem. And the problem was not because of the models. The reason we couldn’t do that was because the data to train the models was not in existence. A lot of our languages across Africa are under-documented, meaning that we do not have them in digitized format. It’s the same as we have a lot of English language texts and French language texts in digitized format, which has been used. And so the data did not exist. And so we started to say, how can we solve this,” Uche said.
He said the team started building the solutions from the ground up, engaging the everyday speakers of local languages to collect data that is truly reflective of how they use the language, how they communicate, the code-mixing they use, the augmentation, the code-switching that they use, etc. That way, the model been developed would be extremely representative.
“And that was how we started. The goal is once we collect that data, that data can really be used to unlock opportunities for them, and serve them in their local language,” he said.
At the fore of the EqualyzAI team are founders who are globally recognised thought leaders both in tech and business.
Dr. Adekanmbi brings extensive experience as a C-level executive at MTN and Airtel Africa UK Limited, where he led innovations, analytics and strategy, serving over 250 million customers across over 20 countries on the continent. He is also a Global Grand Challenge winner for his work on Large Language Models (LLMs) and has played a pivotal role in establishing Data Science Nigeria, Africa’s leading AI learning, research, and solution delivery network.
Dr. Adebara is a globally recognized AI researcher, specializing in Deep Learning, Computational Linguistics and Natural Language Processing (NLP), with a PhD from the University of British Columbia (UBC), Canada. She is a UNESCO IRCAI Top 10 Outstanding Global AI Solutions awardee for her work on AfroLID, Serengeti and Cheetah, groundbreaking models for language identification and natural language understanding across 517 African languages.
Speaking about the solutions, Dr Adebara emphasized the importance of localized AI solutions, stating: “Language is a gateway to knowledge, opportunity, and identity. By building AI that understands and speaks African languages, we are preserving our cultural heritage and creating tools that can transform lives. Africa is home to over 2,000 languages, yet most AI models struggle to understand even a fraction of them. Our goal is to change this by developing AI that speaks and understands Africa’s languages, unlocking new opportunities for education, business, and social impact.”
Technext Newsletter
Get the best of Africa’s daily tech to your inbox – first thing every morning.
Join the community now!
