The problem
Water is Jordan’s binding constraint and irrigation runs on vibes. A Valley farmer runs his drip lines two hours every morning because his neighbor does. On cool cloudy weeks he is pumping expensive groundwater straight past the root zone; on heatwave weeks the peppers stress before noon. The agronomy to fix this — daily crop water need from weather data — has existed for decades but never reached a farmer without a sensor salesman attached.
The build
Two people ship a scheduler, not a sensor. Onboarding is one WhatsApp location pin, the crop, and the drip flow rate. Every night a job pulls the farm’s reference evapotranspiration, soil moisture, and rain forecast, applies the crop’s coefficient for its growth stage, and sends a morning message in Arabic: run 47 minutes today, skip tomorrow, rain Thursday. The farmer replies “done” or “skipped” and the schedule adjusts. The free weather API tier covers roughly ten thousand farm-days of calls per day — the whole pilot runs on zero data cost.
Why now
The missing input — per-field daily ET0 and soil moisture — recently became a free API call instead of a weather station. Open-Meteo exposes reference evapotranspiration, multi-layer soil moisture, and precipitation forecasts free up to 10,000 calls a day. Delivery rides the same July-2025 WhatsApp pricing shift: the farmer’s morning question opens a window and the schedule lands free. Three years ago this needed hardware in the soil; now it needs a location pin.
First customer
Jordan Valley vegetable growers pumping metered groundwater. They feel every wasted cubic meter in dinars, they mostly already run drip, and a single season of trimmed pumping hours covers a subscription several times over. Sell through well-driller and drip-equipment shops that already visit these farms monthly.
Challenges
A schedule is a promise, and soil is not uniform — one farm’s loam holds water two days longer than the neighbor’s sand, so the bot must learn per-farm corrections from “too dry / too wet” replies instead of defending its math. Flood irrigators are out of scope until a version speaks in irrigation turns, not minutes. And in a drought year, telling a farmer to water less when his well is failing reads as blame — the message must frame savings as stretched supply, not lectures.