in , ,

Bridging the Language Gap: EqualyzAI’s Hyperlocal Approach to African AI

EqualyzAI, an artificial intelligence startup focused on democratizing AI for African language speakers, has officially unveiled its suite of solutions. The company aims to build AI technologies that address real-world challenges in emerging markets through locally relevant approaches.

The startup develops its solutions using hyperlocal multimodal datasets created in partnership with native language speakers. This approach enables the creation of localized models for AI innovations that truly understand African contexts and languages, making technology more accessible and impactful for underserved communities.

EqualyzAI has launched three key solutions targeting critical needs across Africa. uLearn serves as an LLM-powered tool helping teachers generate educational materials including lesson plans and quizzes in both English and local languages. DataCollect functions as a platform for gathering the hyperlocal datasets necessary for building inclusive AI solutions supporting numerous African languages. Their third offering, AfroSLM 1.0, represents a Small Language Model optimized specifically for improving financial literacy in Yoruba and Hausa languages.

During the launch event on International Mother Language Day, Co-founder and CEO Dr. Olubayo Adekanmbi emphasized their mission of creating AI that speaks African languages and understands African contexts across multiple formats. He described their work as focusing on empowerment and inclusion by developing AI solutions trained on African realities.

Founded in January 2023 by Dr. Adekanmbi and Dr. Ife Adebara (CTO), EqualyzAI concentrates on building AI solutions for native language speakers throughout Africa. Their approach centers on leveraging hyperlocal data reflecting everyday language usage patterns. The company has developed powerful small language models ready for integration into various products and services to reach last-mile users. Their ambitious goal involves processing over 75 African languages, each spoken by at least one million people.

According to Uche Edwin Okay, Product Development Lead at EqualyzAI, the team identified significant shortcomings in how existing chat models and translation tools handle African indigenous languages. They demonstrated how both ChatGPT and Google Translate failed to provide accurate translations that show contextual understanding of languages like Yoruba, Hausa, and Igbo.

The startup addresses this gap by engaging directly with everyday speakers of local languages to collect authentic data reflecting real communication patterns, including code-mixing and code-switching behaviors. This approach ensures their models achieve high representational accuracy.

EqualyzAI’s founding team brings substantial expertise to the venture. Dr. Adekanmbi previously served as a C-level executive at major telecommunications companies, while Dr. Adebara is recognized for her pioneering work in computational linguistics and natural language processing for African languages, earning a UNESCO award for her models covering 517 African languages.

Dr. Adebara highlighted language as a gateway to knowledge and opportunity, noting that while Africa contains over 2,000 languages, most AI models struggle with even a small fraction of them. Their mission focuses on developing AI that preserves cultural heritage while creating tools that transform lives across educational, business, and social domains.

Source

What do you think?

Written by Sylvia Duruson

Leave a Reply

Your email address will not be published. Required fields are marked *

Digital Banking Revolution: Affinity Africa’s $8M Funding Transforms Financial Access”

Nigerian Fintech Risevest Obtains U.S. Broker-Dealer License, Expanding Global Reach