Rwanda deserves recognition for Miss Baza, an innovative voice-first artificial intelligence assistant launched by MTN Rwanda in partnership with Viamo. It brings AI closer to ordinary citizens, including those without smartphones, mobile internet or mobile applications.
By dialling 845 from an MTN line, users can ask questions and receive spoken responses. The service supports Kinyarwanda, English, French and Kiswahili, covering areas such as healthcare, agriculture, financial literacy, land titles and local government services. Users receive 30 free minutes monthly, after which calls cost RWF 45 per minute.
This is a commendable innovation. It recognizes that access to technology should not be determined by smartphone ownership. However, beyond the impressive experience, there is a bigger technological story worth examining.
What exactly powers Miss Baza, and where does Rwanda stand in owning that technology?
Understanding the technology
When someone calls Miss Baza, several systems work together. MTN provides the telecommunications network and access through short-code 845. Viamo provides the voice platform and interactive voice response infrastructure. Pindo, working with Viamo, contributes to Kinyarwanda speech processing. Google’s Gemini provides the underlying generative AI capabilities, while local information sources, including Eden Care, contribute contextual knowledge relevant to Rwandan services.
The spoken question is converted into information the AI can process. The model generates a response, which is subsequently converted into speech and delivered to the caller. This collaboration brings together local and international capabilities. But contributing to a product is not necessarily the same as owning its underlying technology.
Who owns the intelligence?
Google developed Gemini, the foundational AI model powering the service. Viamo contributes its voice platform and orchestration technology. MTN Rwanda provides the telecommunications infrastructure, distribution and customer access. Pindo contributes local language-processing capabilities.
These are valuable contributions, but they raise important questions about intellectual property.
Does MTN Rwanda own the complete source code? Who owns improvements made to the system? Who controls the language resources developed through the collaboration? Can Rwandan engineers independently modify and reproduce the entire architecture?
The specific contractual arrangements governing Miss Baza’s intellectual property have not been publicly established. Nevertheless, owning a service or its brand does not automatically mean owning the technology behind it.
The question of data
Every conversation with Miss Baza potentially generates valuable information about citizens, their languages, questions, needs and interactions.
Where are voice recordings and transcripts processed and stored? Who can access them? How long are they retained? Are conversations used to improve AI systems? Under what agreements? We should not assume that Google trains its models on these conversations. That requires evidence. But the processing arrangements, data-retention policies and responsibilities of each partner deserve transparency.
There is also an economic question: if local language resources and conversational data contribute to improving AI systems, who ultimately captures the resulting value?
The advantages and dependencies
There are clear advantages to this arrangement. MTN can launch quickly without building a foundational AI model or investing immediately in expensive GPU infrastructure. It benefits from Google’s established AI capabilities and Viamo’s voice technology. Citizens gain access to services through ordinary telephones.
However, dependence carries risks. Changes in API pricing, licensing conditions or international service availability could affect operations. Local engineers can maintain components they control, but they cannot independently repair Google’s proprietary model or Viamo’s closed systems without the necessary access and rights.
This is not an argument against international partnerships. It is an argument for ensuring that partnerships build local technical capacity rather than permanent dependence.
The bigger economic question
Africa has historically exported raw materials and imported finished products. We must ask whether a similar pattern is emerging in artificial intelligence.
Our languages, accents, cultural knowledge and human interactions are valuable resources. They can contribute to AI development. Yet if foundational models, computing infrastructure and intellectual property remain externally controlled, Africa risks becoming a market and source of inputs for technologies whose highest-value assets are owned elsewhere.
Beyond access
Miss Baza is an important achievement, and MTN Rwanda deserves recognition for making AI accessible through ordinary telephone calls.
But technological progress should not be measured only by how many citizens can use artificial intelligence. It should also be measured by how many local engineers can build it, how many companies own its intellectual property, and whether the country can operate and maintain the underlying architecture independently.
James Kaliisa is Co Founder & CTO of Nexus Inc., a Rwandan deep tech company building sovereign artificial intelligence infrastructure and digital systems for Africa.
