By Roman Aglaganov
Plants send electrical signals through their tissues when they need water, light or nutrients, or when they are under attack. Electric Plant, a startup founded by Justin Mast and the neuroscientist Lee von Kraus, has built an AI model to read those signals and tell growers what a plant needs. TechCrunch reported on the company and its funding on 20 October.
The model
Bombadil: the company calls it “the first foundation model for plant electrophysiology”. Trained on three years of data the company collected, it interprets signals from about 125 species, from ferns and redwoods to tomatoes and dahlias, and is meant to work on species it has not seen.
The sensors: probes about the size of a folded phone are screwed into the living layer under the bark and send data to a base station over a low-power wireless network. Each costs tens of dollars, where existing plant monitors cost thousands, and the company says a couple per acre are enough.
The output: the model runs in the cloud, and growers use its readings to adjust watering and fertiliser. Mast claims it can raise yields while cutting water use by 20% or more. The company has not named its first customers, and the claim has not been independently tested.
The money
Electric Plant raised an $11.2 million seed round led by Onto Ventures and Climactic. “We’re listening more than we’re speaking, but listening’s the first step,” Mast said.
Why it matters
Most farm sensors measure the soil or the weather and guess what the plant is experiencing. Reading the plant directly could tell growers about drought stress or pests sooner. It is also a reminder that “foundation model” no longer means language: the same recipe of lots of data and one general model is spreading to signals no human can read.
