Mantle labs offers unclouded vision of global Agri coverage and risk


There was a big drop in the oilseed rape acreage this year, 28 per cent less than initial government estimates. However, one start-up company picked this up early and flagged the issue to its clients. That start-up was Mantle Labs, a leading remote sensing analytics company, which has developed a revolutionary AI algorithm called Helios for ‘seeing through clouds’, increasing the accuracy of satellite imagery for risk assessment and crop monitoring. 

Jon Pierre of Mantle Labs was an agricultural commodities trader for ten years, and saw the potential to improve prediction of yield and management of risk using satellite data – but the issue was clouds in the imagery. He was put in touch with Mantle Labs founders Swapnil Baokar, Rishi Sapre and Prof Clement Atzberger, who had a breakthrough solution to this issue and had created a risk management system for banks, insurance and crop input companies in India to enable them to service farmers of all sizes.

Jon Pierre explains: “Banks in India have to lend about 18% of their loan book to agriculture and famers, but there are 100 million smaller farmers in India with no financial records. The team at Mantle Labs used historical satellite data and AI to analyse land use and provide risk management tools which were previously unavailable. 

“We saw the opportunity to offer a world view of global agriculture with our Geobotanics crop monitoring platform, which mixes data from multiple satellites to provide a daily update with zero interference from clouds.”

Mantle Labs now has a presence in the UK, India and Austria and operates internationally. It recently won the special commendation award at the Financial Times / International Finance Corporation Transformational Business 2020 Awards in the Food, Land and Water Category.

Leading growers in UK and overseas are using Geobotanics for precision agriculture to understand variability across the field and apply inputs – fertiliser, plant protection, water – appropriately. This creates a field level profitability analysis and would enable decision making about future strategy.

Other users of the Geobotanics platform, such as food retailers, crop input providers, banks and insurers want a more macro view to understand trends and impacts – drought impact analysis, global crop performance monitoring. The platform also provides an index for credit risk and insurance pricing, underwriting to claim probability.

“Retailers source produce worldwide, by using a dashboard they can quickly see issues, such as drought or flooding and forecast production,” Jon says.

Land use analysis and crop identification is still often done manually and Geobotanics can provide an accurate and automated solution.

Jon continues: “Covid-19 has made remote monitoring more essential; for one project in Italy we provided an assessment of where and when corn had been planted and if alternatives crops had been sown. They would normally have sent people out but this hadn’t been possible due to the pandemic. Satellite data is probably more accurate and then you have the benefit of extracting further benefit, for example comparison with historical data to enable forecasting of yield. 

“In the UK the government initially projected that oilseed rape was 500,000 hectares this year. Following a poor start of season due to extreme climate events, our early assessment put this at 360,000 hectares, which in the end proved spot on. When combined with poor yields, this has huge implications for the whole value-chain.”

Belinda Clarke, Director of Agri-TechE, says: “Cloud cover has been a rate-limiting factor in the use of satellites in agriculture and Mantle’s algorithm provides a way to overcome that. The ability to offer a world view of agriculture offers fantastic opportunities to predict and monitor land-use and crop productivity anywhere on the planet.”

Mantle Labs is currently working across the entire agri-food value chain to improve forecasting and risk assessment.


About Author

Leave A Reply