While large language models (LLMs) like OpenAI’s ChatGPT and Google’s Gemini are transforming work, education, and communication globally, most remain optimized for English, leaving millions of Africans — particularly speakers of underrepresented languages — excluded from the AI revolution.
Three Ethiopian innovators, Bekalu Temesgen, Baslael Selamu, and Eyob Desta, are working to close that gap with Ras: Ethiopian AI, a multilingual AI assistant trained to understand and respond in Amharic, Tigrinya, and Afan Oromo — three of Ethiopia’s most widely spoken languages.
Bringing Local Languages Into the AI Era
Built on Google’s Gemini model, the Ras team developed a custom fine-tuned layer using open-source datasets and Ethiopian social media content to handle:
- Regional dialects
- Cultural slang
- Informal language structures
“AI tools are becoming more available globally, but a huge portion of Ethiopians are left out because they don’t understand English,” Bekalu explained. “We wanted to build a platform that speaks our languages and reflects our culture.”
Story Builder: AI-Powered Learning for Children
One of Ras’s standout features is its story builder mini app, designed to create interactive storybooks for children in local languages:
- Basic Mode: Users type or speak prompts like “Tell a story about Emperor Tewodros for my 5-year-old daughter”. Within 15–20 minutes, Ras produces a fully narrated, illustrated, and subtitled storybook.
- Advanced Mode: Users can design custom characters or upload story drafts, which Ras transforms into narrated audiobooks complete with visuals.
By localizing both language and storytelling, Ras aims to preserve Ethiopian heritage while making AI accessible to young learners.
Expanding Into Business Guidance and Agentic AI
Beyond storytelling, the team is developing an agentic mini app to help Ethiopians start and formalize businesses. The planned features include:
- Step-by-step business registration guidance
- Automated document generation
- Advice on permits, licensing, and government processes
- Integration of real-time legal and regulatory updates
Bekalu says the goal is to create “an AI-powered guide that walks users through everything from developing a business idea to knowing which government offices to visit and what paperwork to prepare.”
Looking Ahead: Regional Impact and Swahili Expansion
The Ras team plans to expand beyond Ethiopia by adding Swahili, a regional lingua franca spoken by over 100 million people across East Africa.
Their broader vision is to:
- Democratize AI access for underrepresented communities
- Build tools that reflect African contexts and languages
- Enable inclusive participation in the digital economy
Why Ras Matters
Africa’s AI accessibility gap is as much about language as it is about infrastructure. By training Ras to understand cultural nuance, dialects, and local storytelling traditions, the team is showing how homegrown innovation can solve global technology inequities.
“Our aim isn’t just to build another chatbot,” Bekalu says. “We’re creating tools that empower Ethiopians, preserve our stories, and help people participate meaningfully in a world increasingly driven by AI.”
