Plants Speak in Electricity, and a Startup Is Learning to Listen

Plants Speak in Electricity, and a new startup says it has learned to listen. The San Francisco company Electric Plant emerged from stealth in October 2026 with an eleven-point-two-million-dollar seed round led by Onto Ventures and Climactic, money it will use to bring a plant-reading sensor and an artificial intelligence model to market, according to the company's announcement.

The company was founded by Justin Mast, who previously built the online garden retailer Bloomscape, and Lee von Kraus, a neuroscientist who has worked with DARPA and NASA and previously co-founded the neurotechnology company Halo Neuroscience. The pair spent three years collecting plant electrical data before stepping into the open with a sensor called Bud and a foundation model named Bombadil.

The idea began with a cyborg plant. Years ago, Mast was reading scientific literature and came across research on a houseplant wired to a motorized stand: when the plant wanted more light, electrical signals moved through its living tissues, a control board sensed the current, and the stand wheeled the plant closer to the light source. Mast told TechCrunch the discovery sent him down what he called a rabbit hole, and the startup is the result of following it.

Bud attaches to a plant and records its electrical activity. Bombadil then translates those signals into plain guidance — water, nutrients, or attention — replacing costly in-person inspections by growers and agronomists, according to the announcement carried by Business Wire. The founders believe the same approach could one day let people talk back to plants, but for now the company says it is focused on listening first.

How the Bud sensor and Bombadil model work

The training corpus is unusually large for plant science. The model was pretrained without labels on more than four million hours of electrical recordings taken from over one thousand four hundred plants across one hundred species, the company said in a technical whitepaper released alongside the launch. From that base, Bombadil can estimate markers including photosystem II efficiency, leaf temperature relative to the surrounding air, and the Crop Water Stress Index, an established agricultural standard for measuring water stress — even in species it never encountered during training.

That generalization is the part the company's machine-learning team is proudest of. James Kirk, the startup's head of artificial intelligence, said most earlier models of plant electrical signals worked for only one species in one setting, so a new plant or a new environment meant starting over. Because Bombadil learned what is common across species, the company says, it can read unfamiliar plants without retraining — a claim that would let the same sensor work on a greenhouse tomato and a national park pine. It is the same kind of leap that let researchers spend eighteen years decoding the calls of a mystery whale, except here the signal is voltage instead of sound. Eighteen years of whale calls revealed a mystery whale's habitat through patient pattern-finding, and Electric Plant wants to compress that kind of discovery into a single model.

Why growers and gardeners should care

Today's growers mostly read plants from the outside. Farmers, greenhouse operators, and foresters rely on soil moisture probes, handheld spectroscopes, and satellite imagery — coarse tools that react to visible symptoms after stress has already set in, the chief executive said. A direct line into the plant's own signaling could let irrigators act hours or days earlier, shifting plant care from guesswork to measurement.

The water stakes are enormous. Agriculture consumes the largest share of the world's fresh water, and the company claims its approach could cut irrigation by a fifth or more — a figure reported by Notebookcheck that the startup says would matter most in drought-prone regions. When wildfires torched an area twelve times the size of London, the link between heat, dryness, and plant stress became impossible to ignore, and tools that catch thirst early could become standard equipment. For home gardeners, the promise is simpler: a sensor that tells you the plant is thirsty before the leaves droop.

There are reasons to stay skeptical. Independent analysts at explainx.ai published a claimed-versus-verified breakdown of the announcement, noting that the water-savings figure and the sensor's economics are company claims rather than independently tested results. The firm has also declined to name its first customers, and the whitepaper has not been through peer review. As young Canadians take their government to court over its climate record, the pressure on agricultural claims to hold up under scrutiny is only growing. Plants Speak in Electricity only matters if growers trust the translation — and trust will come from field data, not launch-day press.

The company says Bombadil is built to grow with its users: new tasks, from flagging a specific disease to predicting bloom timing, can attach as readouts on the same core model instead of requiring a new model each time. If that platform vision holds, the Bay Area team will have turned a cyborg houseplant into the ears of modern agriculture. For now, the industry is watching to see whether the electrical chatter of all those species turns out to be as legible as the launch suggests. Plants Speak in Electricity — the next step is proving farmers can understand the reply.