Tech Nation chooses Cervest to join Applied AI programme

Being on the programme will fundamentally help us accelerate the delivery of benefit to users.

Cervest’s Earth Science AI platform is designed to transform the decision-making capabilities of businesses, governments and growers in the face of increasing climate volatility. The platform automates the collection and analysis of data, then generates early, accurate and personalised ‘climate signals’, which are uniquely accessed on a price-per-signal basis served over the cloud. This makes it an extremely flexible and economical way for users to receive ‘decision useful’ intelligence without experiencing any of the overhead and complexity.

We are building a pioneering scientific framework, rooted in computational statistics and machine learning, which will make it possible to simultaneously interrogate and process complex scientific and operational data including soil health, weather, water, land-use classification and satellite imagery across different scales, resolutions and timescales for millions of hectares of land. From this we are able to extract personalised signals to forecast land productivity and inform land use over time, supporting mission-critical decision making both in-season and into the future.

These early, accurate and personalised streamed signals (which do not rely on on-the-ground sensors or drones) can be incorporated into core government and business decision processes via a dashboard or API enabling them to be automatically shared across the enterprise and their value chain, allowing other business users to incorporate Cervest’s signal analysis into their respective decisions and processes relating to land.

Our first data product offering is the provision of early, personalised ‘streaming’ signals on 2019 in-season crop yield at multiple spatial scales. This beta product is currently used by food and beverage companies to augment their decisions on volume, pricing, sourcing and related planning for ‘at-risk’ procurement orders, which we aim to deliver at a demonstrably lower cost compared to current means of yield estimation.