Lusaka’s water AI project begins with a data inventory

A new AI project for Lusaka’s water utility will begin with a week spent mapping its data. UNDP Zambia announced the partnership on 14 October, with the first assessment intended to establish what information is available, how reliable it is and which applications it can support.[1]

The work is being commissioned before the proposed platform is built. A UNDP procurement notice published on 9 October seeks a Zambian consulting firm for data mapping, AI readiness, preparation and implementation recommendations. It describes operational information spread across separate systems and a planned platform for bringing it together.[2]

Lusaka Water Supply and Sanitation Company already holds production, meter, billing, customer and network records. The pilot is meant to connect them. Its target, non-revenue water, includes leaks as well as unauthorised connections, inaccurate meters and gaps in billing.[1]

A billing omission and a leaking pipe call for different investigations. Combining records could help distinguish water that escaped from water that reached a customer without being properly accounted for.[1]

The tender sets out a specific job for the proposed system: produce more consistent reporting, identify unusual patterns and help staff target investigations and interventions. An anomaly would therefore be a lead for follow-up, requiring an explanation that could be checked against the utility’s operations and records.[2]

UNDP says the initial sprint will establish a loss baseline where feasible and guide the next stage. Its announcement describes a prototype and says the partnership’s memorandum of understanding is still to be signed. It reports no water savings from the project.[1]

The immediate assignment is to map the utility’s data and recommend what can usefully be built from it. That assessment will help shape which losses the later pilot can investigate.[2]

Report an error or suggest a correction