Plants have been telling farmers what they need for as long as there have been farmers. The trouble is that the messages arrive as wilting, yellowing and stunted growth, which is a bit like learning your car needs oil from the smoke. Electric Plant, a startup that has just raised an $11.2 million seed round, wants growers to hear the request earlier, by reading the faint electrical signals running through a plant's stem.
The cyborg houseplant that started it
The company's origin story begins with a houseplant. Justin Mast, now co-founder and CEO, was reading scientific literature when he came across research on a plant wired to a motorized stand. When the plant wanted more light, electrical signals moved through its living tissue. A control board picked up the current and rolled the plant toward the light.
"I'm like, what the heck?" Mast recalled. He describes that moment as the entry to the rabbit hole.
Mast eventually teamed up with Lee von Kraus, a neuroscientist who has worked with DARPA. Together they founded Electric Plant, and for the past three years the company has been collecting data to build Bombadil, a large language model meant to interpret the electrical signals in a plant's stem and pass details about the plant's needs to growers.
The founders' longer-term ambition is a tool that lets people "speak" to plants. Mast is modest about the current state of things. "We're listening more than we're speaking, but listening's the first step," he said. Right now, then, the product is a very sophisticated pair of ears, not a conversation.
The money
Electric Plant's seed round totals $11.2 million and is meant to take its sensors and Bombadil to market. Onto Ventures and Climactic led it. Other participants were:
- Ponderosa
- Endeavor Catalyst
- Ludlow
- Michigan Rise
- The Fund
- Schox VC
- Dug Song
What is wrong with the current tools
Mast's pitch rests on a claim that will be familiar to anyone who has watched a greenhouse manager squint at a dashboard. Even in the most tightly controlled greenhouses, he says, growers lack full visibility into what their plants require. Farmers, growers and foresters have traditionally relied on coarse tools that react to a plant's demands for water, nutrients or light after the fact.
Soil probes, handheld spectroscopes and satellite imagery all work well enough, in Mast's telling, but they leave plenty to be desired. "That technology is sophisticated, but its ability to get at the inside of the plant isn't very sophisticated," he said. In his view, most of it lags the real-time needs of the plant.
There is a nice irony in the next detail. Electric Plant used those very instruments to train Bombadil. The old tools, in other words, served as the teachers for the model that is supposed to replace them. Mast is confident about how that turns out: "We know the model will surpass a lot of that equipment." That is a statement of belief from the CEO, not a published benchmark, and the report offers no head-to-head numbers.
How the hardware works
The sensors are roughly the size of an iPhone folded lengthwise. They attach to a plant with stainless steel screws, and those screws have to go through the cambium, the layer of living tissue just beneath the bark, in order to pick up the microvolt-scale changes happening inside the plant. So this is not a clip-on gadget. It is a small, deliberate puncture.
The data then takes a short trip:
- Sensor to base station. Readings travel over a low-power wireless network.
- Base station to cloud. Bombadil interprets the signals there.
- Cloud to customer. Results go back to the grower, who can, for example, adjust watering or fertilizer schedules.
Mast says the company needs only a couple of sensors per acre to get accurate results. He also claims that a system like this can boost yields while cutting water use by 20 percent or more. As with the performance claim above, that is the founder's assertion, and no supporting data is given.
Price is the other pitch
Existing tools of this kind generally cost thousands of dollars. Electric Plant's sensors, Mast says, cost "tens of dollars." If that holds at scale, it matters for the "couple of sensors per acre" arithmetic, since cheap hardware is what makes it plausible to leave devices screwed into trees and crops across large plots.
A model with a wide botanical range
Bombadil can currently interpret electrical signals in about 125 species. Mast says the first hundred-plus species cover a wide range of plant forms, from ferns, redwoods and tomatoes to dahlias. That spread is notable because a fern, a redwood and a tomato plant have little in common beyond being green.
Mast is not shy about what that adds up to. "This is the most diverse model ever created on biological electrical signals," he said, adding that no one has done anything equivalent in the animal world. Both are the CEO's characterizations, and the report does not independently verify them.
What we do not know
The report leaves a fair number of gaps, and they are worth stating plainly:
- Customers. Mast would not say who Electric Plant's initial customers are. Agriculture and forestry are the obvious fits, but that is inference, not disclosure.
- Model details. Bombadil is described as a large language model, but nothing is said about its architecture, size or how electrical signals are turned into something a language model can read.
- Evidence. There are no published accuracy figures, no trial results and no independent validation for the yield and water claims.
- Plant effects. Sensors that screw through the bark into living tissue raise an obvious question about long-term effects on the plant. The report does not address it.
The kicker
The idea that plants communicate in electricity is a charming one, and the startup has enough funding to find out whether it is also a business. For now, the plants are talking, the sensors cost about as much as a nice lunch, and the company says the model is listening. Whether growers will pay to hear what a tomato has to say is a question the next few harvests will answer.