Agricultural detail at scale
~90% crop classification accuracy
Crop AI · flagship

From emergence to
harvest, watched.

A live phenology and stress model for arable crops, parcel by parcel, across the UK and EU. Built for processors, growers, and the agronomists between them.

PHENOLOGY · WHEATGS21GS31GS39GS61GS75CURRENT · GS39 BOOTINGNDVI0.90.2
Stress flag · 12 ha
Drought stress detected
−18% NDVI vs catchment
~90%crop classification accuracy
8crop types
GS21–99phenology resolution
6 drevisit
Why a flagship

The model nobody else has.

Crop AI fuses radar backscatter, optical NDVI, soil moisture and medium-range weather into a single phenology model trained on thousands of ground-truth fields. It runs every six days on every parcel — not a sample, not a regional aggregate.

Sense

Phenology, per parcel

GS-stage tracking from emergence to harvest. Wheat, barley, rapeseed, maize, potato — modelled per crop, per parcel, every six days.

Flag

Stress detection, ranked

Drought, disease pressure, lodging risk. Flagged before you'd see it from the road, ranked by severity and area at risk.

Yield

Harvest forecast

T/ha forecasts at the parcel, farm, and catchment. Built for processors planning intake, contracts and exports.

The proof

Calibrated against thousands of ground samples — and validated each season.

We publish a validation set every quarter. Real fields, real harvests, model output filed before the combines moved.