Messium Seed

{"summary_one_liner": "Messium uses daily hyperspectral satellite imaging and AI to measure crop nitrogen and generate variable-rate fertiliser plans.", "detailed_description": "Messium provides precision nitrogen management for arable farming. Its system captures hyperspectral satellite imagery of fields at 160 wavelengths of light, which the company contrasts with the two bands used in conventional NDVI analysis, and delivers analysis at 5-metre field resolution. An AI model trained on 30,000 physical laboratory crop samples interprets the satellite data alongside weather, soil and farm management information to estimate the crop's current nitrogen concentration.\n\nThe output is a set of operational recommendations rather than a raw index: current nitrogen concentration, a forecast of when a field will become nitrogen deficient, and the rate to apply. Recommendations are stated to be optimised for the grower's protein target, wheat variety genetics and soils, with the stated objective of maximising profit rather than yield alone. Recommendations are automatically converted into variable-rate application (VRA) maps exported in ISO-XML, Rx and Shapefile formats for direct loading into sprayers and tractors. The platform also tracks progress against yield and protein targets during the season and reports nitrogen use efficiency, breaking out the share of applied nitrogen taken up by the crop versus that lost to leaching or as nitrous oxide, supported by satellite-based soil assessments each season, to produce an audit-ready nitrogen record.\n\nThe company frames the problem it addresses as widespread mis-fertilisation, citing figures on its site of over half of farmland being incorrectly fertilised, more than £50 per hectare wasted, 2-12% of yield potential lost, and 30% of applied nitrogen lost to groundwater or the atmosphere."}</parameter>

<parameter name="metrics">[{"metric_name": "Spectral bands captured per scan", "value_number": 160, "unit": "wavelengths", "as_of": "2026", "source_idxs": [0]}, {"metric_name": "Field analysis resolution", "value_number": 5, "unit": "metres", "as_of": "2026", "source_idxs": [0]}, {"metric_name": "Physical lab crop samples in AI training dataset", "value_number": 30000, "unit": "samples", "as_of": "2026", "source_idxs": [0]}, {"metric_name": "Stated lab-equivalent accuracy", "value_text": "85%+", "as_of": "2026", "source_idxs": [0]}, {"metric_name": "Update frequency", "value_text": "Daily field updates", "as_of": "2026", "source_idxs": [0]}]