Somewhere between when water leaves the treatment plant and the household tap, Lusaka loses water. This is the result of a variety of causes, including pipe leakages and unmetered connections. All this costs money that could be spent expanding and improving services for Lusaka’s residents. While some of the causes are known, the exact points of loss are still a mystery – one that data can solve.

The United Nations Development Programme (UNDP) in Zambia and the Lusaka Water Supply and Sanitation Company (LWSC) has officially launched the "Advancing Zambia's Water Efficiency with AI-Driven Data Intelligence". The soon to be signed MOU will kick off a series of activities set with a pilot that will use artificial intelligence to help LWSC find, understand and prioritise sources of Non-Revenue Water (NRW) - water that is produced but lost through leaking pipes, unauthorised connections, inaccurate meters or billing gaps.

LWSC already collects a wealth of data on water production, metering, billing, customers, and its distribution network. The pilot will bring these datasets together for the first time in one integrated platform. It will then use AI to read them for patterns that people alone might miss.

Doing will take significant computing power, and that is where timbuktoo EDGE programme comes in. Part of timbuktoo Africa, UNDP’s continent-wide initiative to make Africa’s young innovators the engine of a new innovation economy, EDGE is building a pan-African network of sovereign AI compute infrastructure, starting with an initial cohort of five countries. Instead of sending sensitive data to costly cloud servers managed by external parties, governments, universities and innovators can develop, train and test AI models on home soil, with the data kept under national control. Zambia is one of the first countries in the cohort, and its node is hosted at the University of Zambia AI UniPod, which is where LWSC’s data will be put to work.

The result will be a prototype NRW Decision Intelligence Platform that will help LWSC teams spot the anomalies and flag unusual patterns, map the hotspots, prioritise the response and plan to turn trends into better operational and investment decisions.

The journey will begin with a week Data Readiness Assessment Sprint, during which the team will map LWSC’s datasets, assess data quality and availability, establish an NRW baseline where feasible, and identify priority AI applications based on the findings. The assessment will inform the next steps and guide the implementation of the pilot. Over the course of the pilot, the team will work with LWSC to develop, test and validate AI-driven solutions, with capacity building integrated throughout the process to strengthen internal capabilities and support long-term sustainability.

The partnership also aims to make sure that particularly women, youth and other underrepresented groups play a meaningfully role in the innovation process. This includes helping to shape how problems are defined and how solutions are built.

As the Acting Managing Director, Lusaka Water Supply and Sanitation Company, Eng Kennendy Mayumbelo notes, the pilot supports LWSC's 2023–2027 Strategic Plan, which makes reducing NRW a strategic objective. It also aligns with Zambia's Ninth National Development Plan and its push for evidence-based, technology-enabled public services.

"The ambition is not simply to introduce AI. It is to use AI and data intelligence to support better decisions. Together, we have an opportunity to show what locally relevant AI for public good looks like in practice. As we begin this journey, let us move from data to insight, from insight to action, and from innovation to impact." – notes Chali Selisho, Head of Unit, Inclusive and Sustainable Growth, UNDP Zambia