In clinics across the Northern Cape and Nigeria, we kept hearing the same thing: Healthcare workers weren’t unsure of what to do. They were struggling with when to do it and how to fit it all in.
That insight stayed with me. It shifted how I thought about AI in public health — not as a way to tell healthcare workers what to do, but as a tool to help them decide how and when to do what they already know.
Over the past few months, we spent time deeply understanding the needs and motivations of potential Lesedi AI users across South Africa and Nigeria. Lesedi AI is a recommendation engine that uses a program’s own data to support program management staff, clinical mentors, and decision-makers at various levels to:
- Identify gaps: What target did we not meet , and why?
- Recommend action: What should we do differently to meet this target next time?
- Monitor change: Are there improvements in outcomes after the change?
While co-developing a tool designed to shift healthcare practices, we learnt a few surprising things:
