Why We Pivoted from Africa to the United States
A few years ago, Hurone was deploying cancer navigation technology in Africa.The need was obvious. The patients were grateful. The clinicians were overwhelmed and desperate for support, and governments were excited to partner with us. On paper, it felt like the right place to start. And in many ways, it was. But building AI infrastructure teaches you something quickly: software doesn’t scale in isolation. It scales on a complex web of systems. And the system matters more than the value the software creates, especially in healthcare. When I served on the U.S President’s Cancer Panel in 2024, focused on cancer navigation technologies, I had a rare vantage point. We spoke with hundreds of cancer center executives, researchers, patient navigators, patient advocates, payers, and technology leaders across the United States. One theme surfaced repeatedly was that one of the hardest problems to solve in cancer care is what happens between diagnosis, up until survivorship. Cancer care in America is sophisticated, but also deeply fragmented. Patients move between surgery, infusion, radiation, imaging, labs, financial counseling, survivorship clinics, and community care. Each handoff introduces friction. Each friction point creates risk. Navigation teams try to hold it together. In-basket message volumes explode. Patients send anxious messages at midnight, social barriers derail treatment adherence. Clinicians lack visibility into what happened between visits, making each clinic visit an uphill task to recall all that may have happened from their last visit. And it is precisely this complexity that durable software is built. Lesson 1: Clinical Adaptation Is Possible. Infrastructure Is Harder. Despite the gaps in these systems, we adapted to low‑connectivity environments, device constraints, and learned to operate across uneven terrain. One constraint kept resurfacing: without digital medical records, data infrastructure cannot support compounding AI, and longitudinal intelligence cannot exist. Without interoperable systems, automation is fragile, and without compounding data to improve your foundational model(s), there is no long‑term moat. The United States for example passed the HITECH Act in 2009. That single policy accelerated electronic health record (EHR) adoption nationwide. Even then, it took nearly a decade for digital records to become as widespread as it is today. That kind of systemic shift doesn’t happen overnight. Across many African health systems, digitization is still uneven. Some centers are advanced. Others rely on handwritten notes that can’t be easily read, let alone analyzed. We could build intelligent software. But the underlying substrate wasn’t consistently there yet. Lesson 2: AI Scales on Business Models that Work There’s a second truth we encountered. Healthcare systems in many African markets depend heavily on aid flows and external funding. Global health programs are grant-backed, and the donors often cling to not-for profit models to back, while avoiding for-profit models that show clear paths to sustainability, in our case our situation was compounded by the fact that because we were trying to build around a venture-like scale model, such funding became completely out of the picture. This long-term dependency that aid without transitions to sustainable business models, in my opinion, is one of the major reasons global health is broken in Africa. Digital tools frequently arrive through philanthropic partnerships. That model creates access, which may sometimes be short-lived for a nice peer-reviewed publication. It does not create recurring enterprise software budgets or revenue. Dambisa Moyo, the renowned Oxford-trained economist captures this perfectly in her 2010 book “Dead Aid”. When clinicians or key decision stakeholders loved our product, the question was rarely: “How fast can we deploy across the system?” It was: “Can this be funded externally?” Long-term scale requires procurement muscle — B2B or B2G, or B2B2C unit economics where institutions budget for technology on a business model that improves operational, financial and patient outcomes. In the U.S., that muscle exists. Hospitals absorb risk under value-based oncology models. Payers look for utilization reduction. CMS reimbursement codes are evolving almost as fast as technology evolves. When AI reduces workload or avoids readmissions, someone benefits financially. That alignment changes everything. Lesson 3: Data Infrastructure For Healthcare is Beyond Critical At the same time, something else was happening in the U.S. Ambient AI scribes became some of the fastest-growing health AI companies in 2025. Because documentation burden was enormous.And because health systems already had digital infrastructure to integrate into. Africa doesn’t need to replicate the U.S. journey step by step. It can leapfrog. There are promising companies building voice-first clinical documentation and digitization tools in emerging markets right now. But leapfrogging still requires cultural and economic shifts.Moving from handwritten notes requires budget allocation, workflow redesign, and institutional commitment. The faster adopters tend to be hospitals that already had typed EHRs and some history of investing in software. Others remain constrained by structural economics. The Pivot Was a Systems Decision We didn’t pivot because the need disappeared. We pivoted because we understood and learned from our sojourn in Africa about what it takes to build durable AI infrastructure, and with all the forces aligning in the US from the evolution of the FDA clarity on software as a medical and non-device, CMS aligning reimbursement codes and value-based oncology models, Institutional AI governance frameworks forming, and policy mandates around interoperability, and more importantly the collective American spirit to win and dominate the AI race, I knew it was the right time. The clear question for us now became to build the engine first and scale. My team all understood this and we moved with the speed and execution needed to be successful in a field that is moving at unprecedented speed. What We’re Building Now We are embedding Hurone inside major cancer centers to strengthen end‑to‑end cancer navigation.We reduce clinician burden, surface the right patients at the right time, and turn real‑world signals into structured inputs that fit existing workflows. Over time, this creates compounding journey data that becomes exportable infrastructure. The long‑term vision remains unchanged: deliver world‑class cancer care everywhere. Building in the United States is not a departure from that vision; it is a sequencing decision to prove what’s possible on top of mature
