African Languages Lab launches Mansa, a multilingual, multimodal AI built for Africa

African Languages Lab, the AI research and deployment company behind the largest collection of African language data in existence, today launched Mansa, a multilingual, multimodal AI system built for African languages, accents and contexts. Mansa is available now on the web and mobile, with an API for developers and businesses.

African languages account for less than 1% of the data used to train modern AI. In Common Crawl, one of the industry's largest training sources, English makes up over 40% while all 24 African languages tracked, including Swahili, Hausa, Yoruba and Zulu, combine to less than 0.05%. The result is familiar to anyone working on the continent: models that trip over African accents, misread idioms, lose cultural context, and cost African users more to run.

Compute and talent are the easy part. Data is the bottleneck. Much of Africa's knowledge was never written down. It is spoken. So we spent a decade going out and collecting it, language by language, community by community.Sheriff Issaka, founder and head of research at African Languages Lab

That work produced the largest collection of African language data in existence, spanning more than 70 languages. Thirty of them are in production in Mansa, built on over 100 billion curated tokens and more than 19,000 hours of expert-validated speech, gathered through open-licensed sources, direct community partnerships, and All Voices, the Lab's platform where native speakers contribute and validate data. The underlying research has been peer-reviewed and award-recognised at ACL, one of the field's leading venues.

Mansa deeply understands over thirty African languages, including Twi, Hausa, Ewe, Swahili, Igbo, Amharic and Zulu, alongside English, French, Spanish and Chinese. It handles regional dialects and the natural code-switching people actually use.

What is live today

• •Mansa Chat. An AI thinking partner for learning, writing, brainstorming, coding, and problem-solving in more than 30 African languages and major world languages.
• •Mansa Agent. An autonomous assistant that browses live information, connects to business tools, remembers context across sessions, and runs multi-step workflows on a schedule.
• •Mansa Translate. Contextualized translation that preserves meaning and cultural nuance, not just words.
• •Mansa Transcribe. Speech-to-text calibrated for regional African accents, dialects, and code-switching.
• •Mansa TTS. Natural, expressive, production-ready voices across African languages.
• •Mansa Interpret. Real-time interpretation between two speakers, listening in one language and speaking back in another.

The Mansa API is available at launch, allowing developers, startups and enterprises to build these capabilities directly into their own products and workflows.

For everyday users, that means asking a question in Twi and getting a real answer back, a student working through homework in the language they think in, a voice note turned into a clean written message, or a doctor's instructions interpreted in real time while you sit in the room.

For businesses, a bank or telco can run voice support in Hausa or Twi rather than pushing customers into English. A health organisation can deliver guidance by speech to people who may not read or write it. A newsroom can transcribe an interview conducted in African languages.

The use cases are endless. The Lab built the foundation, and the exciting part is what gets built on top of it. As Sheriff Issaka often puts it: “It is not about what Mansa can do for you. It is about what you can do with Mansa.”

Mansa is available today at mymansa.ai. Developers can request API access at the same address.

What to know: Mansa product & developer FAQs

Mansa products

Mansa AI

Mansa Agent

Mansa Translate

Mansa Transcribe

Mansa Interpreter

API & integration

Data, privacy & security

Language support

Performance & accuracy

Mansa Agent: safety & approvals

Getting started

Important use & safety notice

Mansa is an AI system and may generate inaccurate, incomplete, or inappropriate outputs. Users and organizations remain responsible for reviewing outputs and determining whether they are suitable for their intended purpose.

For healthcare, legal, financial, regulatory, security, emergency, or other high-stakes applications, Mansa should be used with appropriate human oversight, validation, and escalation procedures.

Product capabilities, supported languages, integrations, and availability may change as Mansa continues to evolve.