Electric Plant Company debuts plant-health AI alongside $11.2 million seed round

Electric Plant Company emerged from stealth on October 20, 2026, with $11.2 million in seed funding and an AI model intended to read plant health from electrical activity. Called Bombadil, the model estimates stress markers from plant-mounted sensors, the company says, offering growers a reading from the plant itself rather than its surroundings.

Building a model across 100 species

The company began in 2023, co-founded by CEO Justin Mast and Chief Scientist Lee von Kraus. Mast previously founded and led online plant retailer Bloomscape; von Kraus is a neuroscientist. Their new venture combines hardware that captures electrical activity in plant tissue with software designed to turn those recordings into health information.

For Bombadil’s initial training, Electric Plant says it used more than 4 million hours of electrical recordings from over 1,400 plants across 100 species. The recordings were unlabeled: the model learned patterns in the signals through self-supervised training, rather than starting with a human-supplied description of every recording.

The company says its collection has since grown to 7 million hours. That is the size of its accumulated dataset, not the amount used in the first training run. The initial recordings covered a range of environments as well as different species.

The central claim is that learning across plants lets Bombadil estimate health markers for species absent from its pretraining. James Kirk, Electric Plant’s head of AI/ML, contrasts that approach with earlier models that, he says, worked for one species in one setting.

Bombadil learns what is common across species, which is why it can read a plant it never saw in training. That generalization is the breakthrough, and it is what lets the technology scale.

James Kirk, head of AI/ML at Electric Plant Company

Turning voltage into stress estimates

The process starts with Bud, a sensor that attaches directly to a plant and records electrical signals in its tissue. Those readings go to Bombadil, which converts the stream of raw voltage measurements into estimates of the plant’s condition and presents the results to a user.

Electric Plant says the model estimates seven physiological stress markers from electrical signals alone. These include leaf temperature relative to the surrounding air and the Crop Water Stress Index, an agricultural measure of water stress. The proposed distinction is the source of the information: electrical activity inside the plant, rather than soil moisture, satellite images or a visual inspection.

Electric Plant also released a technical whitepaper with the debut. Its launch announcement describes the training approach and names several estimated markers, but gives no accuracy figures for those estimates. The claims about reading unfamiliar species and translating electrical signals into useful health information remain company-reported capabilities.

The model is designed to support additional tasks without replacing its core. Electric Plant describes future disease alerts or bloom-timing predictions as new readouts attached to that base model. Those examples explain its intended development path; they are not presented as demonstrated features of the seven-marker system introduced at launch.

Plant-care operators are the customers

The system already has a field presence, according to Electric Plant: early deployments are running at commercial sites, with hundreds of sensors installed. The company licenses the technology to existing plant-care operators rather than seeking to replace their businesses, and is accepting additional founding operators as it expands crop by crop.

Onto Ventures and Climactic led the seed round announced alongside Bombadil. Other participants included Ponderosa, Endeavor Catalyst, Ludlow, Michigan Rise, The Fund, Schox VC and Dug Song. The company’s stated next step is to bring more operators into those deployments while extending the technology across crops.