Decoding Plant Signals: Startup Uses AI to Understand Plant Needs

Justin Mast's fascination with plant communication started with an extraordinary discovery—a cyborg plant capable of moving towards sunlight by sending electrical signals. This intriguing concept led him down a rabbit hole of possibilities in the world of plant biology.

A New Kind of Communication

Initially taken aback by the idea, Mast teamed up with Lee von Kraus, a seasoned neuroscientist with DARPA experience, to co-found Electric Plant. Over the last three years, they have developed a sophisticated language model named Bombadil, designed to decode the electrical signals emitted by plants. This groundbreaking tool aims to enable growers to better understand their plants' needs.

Mast, who serves as the CEO of Electric Plant, remarked, "We're listening more than we're speaking, but listening’s the first step." This approach highlights their goal of fostering a dialogue with plants rather than merely reacting to their environmental conditions.

Funding for the Future

To expedite the development of Bombadil and its accompanying sensors, Electric Plant recently secured $11.2 million in seed funding. This round was spearheaded by Onto Ventures and Climactic, along with contributions from various investors, including Ponderosa and Endeavor Catalyst.

Mast explained that even in highly controlled environments, growers often lack precise insights into what their plants require. Traditional methods rely on dated technologies that react to visible signs of plant distress, such as needs for water and nutrients, but these do not address the internal signals that plants communicate.

Innovative Sensors for Accurate Results

Electric Plant's Bombadil model has been trained using data collected from existing agricultural tools, which, while effective, don't provide the depth of insight these plants can convey. "Most of that technology lags quite a bit behind the real-time needs of the plant," Mast commented.

In contrast, the startup's new sensors are both affordable and effective. Priced at just tens of dollars, these sensors are significantly cheaper than traditional tools, which often run into the thousands. Each sensor, about the size of a folded iPhone, is affixed to the plant with stainless steel screws that penetrate the cambium layer, allowing them to detect minute electrical changes.

The data from these sensors is transmitted over a low-power wireless network to a base station. Remarkably, only a couple of sensors per acre are sufficient to yield accurate data, which Bombadil processes in the cloud. This enables growers to make informed decisions about watering and fertilizer applications, with the potential to enhance crop yields while reducing water usage by 20% or more.

Wide Applications Across Plant Species

While specific details about Electric Plant's initial customers remain undisclosed, the potential applications in agriculture and forestry are immense. The model has been initially trained on over a hundred species, spanning a variety of plants from ferns and redwoods to tomatoes and dahlias.

Mast claims, "This is the most diverse model ever created on biological electrical signals," asserting that nothing comparable has been accomplished in the field of animal biology. The ability to interpret these signals opens new avenues for optimizing plant care and agricultural productivity, setting the stage for a revolutionary approach to understanding the natural world.