Fewer than 5% of Sub-Saharan Africa's over 2,000 languages have the digital resources needed to train modern language-AI systems, leaving most African-language speakers underserved by tools like voice assistants and translation apps. Lelapa AI's Vulavula platform offers speech transcription, translation, and sentiment analysis for local languages, and its founders — including CTO Jade Abbott — also co-founded Masakhane, a pan-African research collective that has helped build open-source NLP models for dozens of African languages.
Source: Lelapa AI
Frequently Asked Questions
How did Lelapa AI and its approach to African-language AI come about?
Its founders had already spent years working through Masakhane, a research community started in 2017 to close the gap in African-language NLP research, before turning that grassroots work into a commercial venture in 2022 focused on practical tools businesses could use.
Why do African languages need dedicated AI tools instead of relying on existing global systems?
Most large language and speech models are trained overwhelmingly on English and a handful of other high-resource languages, so they perform poorly or not at all on African languages, which locks hundreds of millions of speakers out of AI-powered services unless dedicated data and models are built for their languages.
How is this technology being used today?
Businesses use Lelapa AI's tools for customer service and communication in local languages, and the underlying research has fed into broader efforts — including a 2026 partnership with Google — to expand openly available African-language speech datasets for other developers to build on.