Kaelo
/ka.ˈe.lo/·Setswana·"to care for, to nurture, to look after"
Botswana imports nearly all of its medicines. Getting them from a port to a patient in a rural clinic, on time and unexpired, is a logistics problem before it is a clinical one.
Kaelo exists to help close that gap.
How a tool like this gets built matters as much as whether it works, so we have written down where we stand: a note on parachute research.
25 August 2025
The emergency
Botswana declared a public health emergency. Clinics were out of medicines for blood pressure, diabetes, tuberculosis and cancer. They were also out of sutures and dressings. Elective surgeries were postponed.
Diamond revenue had fallen and foreign aid had been cut, and the strain eventually reached the suppliers. The government moved quickly once it did. Emergency funding was released within weeks, and distribution was rebuilt so the most remote clinics were reached first, which is not how shortages usually go.
Kaelo is part of the response. It takes the questions the shortage made unavoidable and puts numbers on them: which facilities run out first, how long a stockout lasts once it starts, what it costs to move stock late instead of early.
Built with local expertise, so the next shock meets a system that already saw it coming.
What we do
Kaelo is an antimicrobial supply chain optimization platform developed in collaboration with ACHAP and Botswana's Ministry of Health and Central Medical Stores. It uses advanced mathematical models, including robust optimization and adaptive (affine) decision rules, to determine how medicines should be procured, stored, and distributed across the country's 619 health facilities.
The system accounts for demand uncertainty, transport constraints, procurement costs, and the complex tiered network of warehouses, hospitals, clinics, and health posts that cover 2.4M people across 18 health districts.
How it works
Robust optimization
Instead of assuming a single forecast is correct, the model optimizes against theworst-case demand scenario within a plausible uncertainty set. Decisions that hold up under the hardest month hold up under every month.
Affine decision rules
Rather than fixing every order up front, Kaelo learns a linear policy: "order x if we've observed demand y." This keeps the math tractable while letting the system adapt as real demand unfolds month by month.
Software
A public build of the planning tool, with the optimization model intact and every procurement figure stripped out. You supply the demand and cost numbers yourself.
Open the software →Best on a desktop browser for now. The solver sleeps when idle, so the first load can take a moment to wake up.
Team & collaborators
- Dr. Bartolomeo Stellato (advisor)
- Mr. Lesego Busang
- Mr. Stanley Mapiki
- Mr. Tiro Molefe
- Dr. Khumo Seipone
- Ministry of Health, Botswana
- Central Medical Stores
Global health carries a long record of parachute research: arriving from outside, collecting what is needed, and leaving little behind. Kaelo is structured against that. The questions are set by the institutions that run the supply chain, the data remains theirs, and the software is built to be operated locally.