This Google X spinout wants to make plants programmable - with AI

Image: Heritable Agriculture

February 7 | Steve Ranger

The latest spinout from Google's X division is a sustainable agriculture startup, which has developed an AI platform it says can make plants programmable - with the goal of creating higher yields, improved nutrition, and better resilience to pathogens and climate stress.

Heritable Agriculture was a project originally incubated inside Google’s X, the tech giant's so-called ‘moonshot factory’ which aims to launch technologies that can have a significant impact on the world’s most intractable problems. Heritable is now an independent company focused on advancing plant breeding and biotechnology with funding from FTW Ventures, Mythos Ventures, and SVG Ventures.

Heritable CEO Brad Zamft said that agriculture has been able to accommodate the world’s growing population because of the “beautiful” nature of plants. “Plants are solar powered, carbon negative, self-assembling machines that feed on sunlight and water. It’s hard to imagine a technology that would be cheaper to scale,” he said.

However, there’s a problem: existing agricultural practices are also damaging the environment; agriculture accounts for nearly half of habitable land use, a quarter of anthropogenic greenhouse gas emissions, and 70 percent of the world’s groundwater withdrawals.

Zamft (pictured) said although there is lots of data about plant breeding much of it is messy and hard to parse, which led the team to create a machine learning platform to better understand how plants grow, capturing information across multiple crops and tissues. This helps the company to identify and understand the function of specific genomes in plants. “By understanding those genomes, the crops can then be bred with climate-friendly traits for increased yields, lower water requirements, and higher carbon storage capacity in roots and soil,” he said.

Validating the models involved growing thousands of plants at X inside a growth chamber fitted which took hourly pictures of test plants to measure them as they grew. This allowed Heritable to track the accuracy of their predictions when a plant might flower, and understand how specific changes might affect that budding time.

“From our tests, we see a path towards using our technology platform to identify and predict genetic determinants of plant traits like colour, taste, water requirements, nitrogen requirements, and even the ability to absorb carbon from the air,” he said.

Heritable said it working with tree seedlings company ArborGen, on developing of superior loblolly pine trees. Initial research suggests Heritable's AI can halve the time needed for tree improvement, it said.

More on agriculture and sustainability

Get the more science and innovation news every month in Chemistry & Industry magazine. You can subscribe to C&I here.
Show me news from
All themes
from
All categories
by
All years
search by