It’s Just the Beginning

By Karen Byrnes, Director of Partnerships

Too many of us are suffering. Recent estimates suggest that the number of people facing acute hunger around the world could double this year relative to last year. This dramatic increase is being brought on by the combined effects of COVID-19 and a range of other existing and novel threats, including the legacy of conflict in some regions and new threats from locust swarms in others.

Things are changing so quickly, and across so many regions, that understanding exactly what’s going on on the ground and who is in most need of assistance is extremely challenging. This is a crisis that requires a new set of tools: precise agricultural forecasting to understand where food insecurity will be most acute in coming months and granular measures of economic livelihoods that can anticipate and track who is most at risk. We need to equip our decision makers with data analytics of the highest possible accuracy.

Where We Come In

Our work at Atlas AI is one effort to provide this much needed insight. Born from academic research at Stanford University, Atlas AI’s machine learning algorithms combine terabytes of satellite imagery and other geospatial assets with a rich array of ground-truth data to generate localized insights into agricultural productivity, infrastructure quality, and economic well-being.

“We need to equip our decision makers with data analytics of the highest possible accuracy.”

Through a host of academic research, we’ve shown that a new wave of satellite information from both public and private sensors can be used to make accurate predictions at the field to national scale. We carefully validate our models using ground data collected by ourselves and our partners, and our methodology is backed by rigorous, peer-reviewed scientific research.

We provide data for different time periods back to the early 2000s and also calculate how data has morphed and shifted over time. This data is generated by applying our trained models to imagery from these earlier years, in order to be used to build an understanding of long-term trends.

In early 2020, our work was recognized with the validation of our Series A. Airbus Ventures led the investment, in partnership with Micron Ventures and The Rockefeller Foundation, who’s been a supporter since the earliest days of the company. Our unique combination of investors provides us with an exciting opportunity to develop a business model that thoughtfully integrates profit with public good: a charge that we do not take lightly.

Now on the road to Series B, we are moving full speed ahead, teaming with collaborators from the World Bank, DARPA, the Alliance for a Green Revolution (AGRA), One Acre Fund, McKinsey ACRE, and others, thinking about how we can make data accessible at scale.

Our Commitment to Kenya

Over the past few months, we have been working hard as partners in the Kenyan Ministry of Agriculture’s Food Security War Room. This project is a multi-sectoral initiative to ensure availability, accessibility, and affordability of food and water to Kenya’s impacted populations. We’re doing our part by providing the Ministry with sub-national maize yield estimates back to 2016.

“Our unique combination of investors provides us with an exciting opportunity to develop a business model that thoughtfully integrates profit with public good: a charge that we do not take lightly.”

Atlas AI’s maize yield displays the estimated production of maize grain in a field per unit area. We currently estimate yields at 10m resolution for all pixels deemed to be growing maize in a given growing season. The underlying data supporting these insights come from Sentinel-1 and Sentinel-2 satellite imagery, various gridded weather and soil datasets, and detailed maps of typical cropping practices, such as sowing dates and fertilizer rates throughout the region. Our estimates have been validated against a variety of ground sources, including hundreds of crop cuts acquired over the past five years by One Acre Fund.

Next Steps

Beyond the War Room, we are committed to systems-strengthening and capacity-building. We are working closely with the Kenya Agricultural & Livestock Research Organization (KALRO) and MercyCorps Agrifin to explore new ways of enhancing the government’s capacity to house technology like ours. We are leveraging our academic roots to create tools that allow decision makers to learn about machine learning and geospatial analytics directly from our world-class experts. We continue to work hand in hand with our collaborators in the World Bank’s 50x2030 initiative to close the agricultural data gap, and we are actively seeking new partnerships to help equip decision makers in emerging and global markets.

We are just at the beginning of our journey to mobilize leaders with agronomic and economic intelligence —and to use it to unlock new potential.

Interested in learning more? Contact partnerships@atlasai.us.