S4D. Insights More insights →
01 Cognivo · Rwanda & Zambia · Decision intelligence · In development

A supply chain doesn't need another dashboard. It needs a decision.

Why S4D is building decision intelligence — not another reporting screen — for public-health supply chains in Rwanda and Zambia. Cognivo is in early design and MVP development.

Status key: Realized In design Target

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.

RAW SIGNALS Stock on hand Consumption rate Lead-time variance Cognivo DECISION-INTELLIGENCE LAYER TARGET OUTPUT Recommended action
Central store trends toward an ACT stockout in ~6 weeks at current burn. Reallocate surplus from two over-supplied districts before the next cycle closes.
EXPLAINABLE · RULE-BASED · AUDITABLE
Illustrative target output. The MVP under development is being built to turn fragmented operational signals into one explainable recommendation — not another screen to interpret.
86 features
Data Readiness assessment and an 86-feature engineering specification delivered for the Rwanda MVP — design artifacts, not a live system
Realized
Rule-based MVP
First MVP phase being built as an explainable rule-based narrative layer, over an opaque generative prototype
In design
Owned, in-country
Decision intelligence the Rwanda and Zambia Ministries will own — the destination the build is working toward
Target

More from S4D Insights

S4D Consulting

We build systems that get used — and capability that stays.

Digital transformation and applied AI for public health and the public good across Sub-Saharan Africa. From national health intelligence to the platforms running schools and buildings, the through-line is the same: turn data into decisions people trust, and leave the capability in local hands.

Talk to S4D →