Early-stage startups highlight gaps new AI agency could solve

Emerging software developers and tech startups are calling for infrastructure enhancements and expanded private-sector opportunities to accelerate tech commercialization. The call comes as early-stage innovators navigate key operational hurdles following the government's approval to establish the National Artificial Intelligence Agency. Samuel Mizero, 26, an software developer, welcomed the creation of the new agency, saying it was timely. Mizero is the founder and Managing Director of UMVA AI, a startup that gives young coders access to AI-powered Integrated Development Environments (IDEs)—software workspaces that embed AI directly into the coding process.. We support developers to secure their own data for safety purposes, Mizero said. ALSO READ: Rwanda’s AI Agency is timely, but delivery will matter most He emphasized that as the AI agency needs to address the core challenges facing early-stage companies. We have been working without a central government umbrella regulating AI technology, Mizero said. We expect the new agency to advocate for us to reduce high electricity costs—since AI labs consume significant power—and to push young coders onto the international scene through global conferences and training programs. Mizero added that with stronger public sector collaboration, local startups can help combat online misinformation while building secure domestic data systems. Many people post incorrect information about our country online, Mizero said. Through local platforms and collaboration with startups, the country can ensure accurate information on the internet. He noted that UMVA AI operates servers allowing Rwandans to store data locally, though scaling his AI model will require additional funding and building key partnerships. This directly addresses critical cyber security needs, he added. Data and infrastructure bottlenecks Infrastructure limitations also constrain Incuti AI, a locally developed language model. The tool aims to bridge language barriers by converting handwritten text into digital formats for Kinyarwanda speakers who do not understand foreign languages. However, Incuti AI faces access restrictions when attempting to digitize physical records held by some public and private institutions. They often say documents are confidential and hesitate to collaborate or share access to hard-copy texts, leaving our requests on hold, said Jean Marc Ntirushwa, founder of Circuit Notion, which developed Incuti AI. It is commendable that Rwanda is positioning itself as a continental technology hub, Ntirushwa added. However, there is currently a shortage of infrastructure. We urgently need domestic data centers, expanded funding, and deeper collaboration with AI startups. Despite these hurdles, locally developed AI tools are already delivering results on the ground. Fabrice Nahimana, 28, a secondary school teacher in Nyagatare District, uses local AI platforms to expand his programming skills and streamline classroom work. The local models are far more affordable than international alternatives, Nahimana said, noting that UMVA AI has helped him improve his coding skills. The local models can support school management systems, help teachers prepare assignments, and assist in organizing tutorials. National AI Agency to create opportunities The Ministry of ICT and Innovation (MINICT) acknowledged that early-stage startups face major barriers, including limited access to quality datasets, specialized expertise, computing power, and clear institutional pathways. Esther Kunda, Director General of Innovation & Emerging Technologies at the ministry, assured that the new agency is designed to coordinate these crucial elements. The agency will coordinate and facilitate access to computing capacity that can be used by researchers, startups, and innovators alike, Kunda said in an interview with The New Times. Access to high-performance computing is currently one of the biggest constraints for small companies trying to develop and test AI products. Kunda emphasized that the agency will connect priority public-sector challenges with local innovators, creating structured pathways for startups to test and deploy solutions at scale. Importantly, the agency is not intended to compete with startups by building everything itself, Kunda clarified. Its role is to set priorities, provide platforms and infrastructure, bring partners together, and create opportunities for private-sector companies to build and commercialise solutions. ALSO READ: Inside the newly approved National AI Agency Addressing high overhead costs, such as GPUs, high-performance compute clusters, large datasets, and secure storage, Kunda noted that shared infrastructure will be central to the government's support. Shared compute and data-center capacity will significantly reduce entry costs, allowing many users to leverage expensive resources rather than each company bearing the full capital and operating burden alone, she said. Kunda added that the agency’s goal is to have a stronger and coordinated ecosystem in which startups have customers, researchers have infrastructure and government has access to solutions.

News

Early-stage startups highlight gaps new AI agency could solve