Example Scenarios
ILLUSTRATIVEIllustrative users, not real case studies: who could benefit, and what would change
The problem
She irrigates her kale on a fixed schedule. In a dry week her crop still stresses; in a wet week she wastes water and fuel pumping from the river.
Conduit air temperature every 15 minutes, rainfall totals for the last 7 and 30 days, and regional soil moisture from ERA5.
AFYA MAZINGIRA computes reference evapotranspiration (Hargreaves) and crop water demand (FAO-56 Kc), then balances it against recent rainfall and soil moisture.
"Hold, rain expected" instead of irrigating, or an exact application depth in mm when the root zone is depleted.
Water and pumping fuel would be spent only when the crop needs them, and irrigation would come before stress rather than after visible wilting.
Try this yourself
Open the actual feature used in this story.
How this scales
AFYA MAZINGIRA scales by adding ground stations, not by stretching one station across a region. Each new Conduit station becomes another high-frequency intelligence node inside the same regional outlook, and the capabilities shown here, states, forecasts, Best-Time, farm advisory and bilingual explanation, are built to travel with it. They do not arrive ready-made: the forecast and state models are fitted to this station's own archive, and parts of the code still carry this station's coordinates, so a new location needs its models refitted on its own record before its numbers mean anything.