Public-health supply chains in the region are not short of data. Logistics management systems already record stock on hand, consumption, and lead times across thousands of facilities. Yet stockouts persist — not because the numbers are missing, but because a dashboard hands the analytical work straight back to people who are already stretched thin. Seeing a red cell is not the same as knowing what to do about it before the next order cycle closes.
Cognivo is S4D's response to that gap — and it is being built now. The work is in early design and MVP development; this is the thinking behind it, not a live product. The intent is a decision-intelligence layer that will sit above existing supply systems and produce owned, plain-language recommendations a planner can act on. The emphasis is deliberate: not another reporting screen, but a layer designed to do the reasoning and explain itself.
The hard part isn't the model. It's trust, ownership, and an answer a planner can defend.
In Rwanda, under the RMS AI engagement, that conviction is shaping how the work is sequenced. Before any modeling, S4D delivered a formal Data Readiness assessment — establishing what the data could and could not yet support. The team then specified 86 engineered features across ten families to give the system something meaningful to reason over. Most tellingly, the first MVP phase is being built as a rule-based narrative layer rather than a black-box generative prototype. In a national health system, a recommendation has to be explainable, reproducible, and defensible to the people accountable for it — and a transparent rule base earns that trust faster than an opaque one.
In Zambia, the same philosophy is being designed into the national demand-planning architecture from the outset, so that intelligence is a property of the system rather than a layer bolted on later. The framing S4D carries into both countries is consistent: owned decision intelligence — capability the Ministry will keep — over a vendor dashboard it merely rents.