01
Executive summary
An American farm is a business that runs on records it never connects. Utility bills, pump logs, spray records, payroll, agronomy spreadsheets, and water meters each live in their own silo, and the physical state of the field is barely recorded at all. Problems surface as bills and yield losses, weeks or months after the moment they could have been fixed.
At the same time, the regulatory load on a farm has grown into a second full-time job. Groundwater, pesticides, nitrogen, labor, and food safety each report to a different agency on a different calendar, with fines for every miss. The reporting is data work, performed by hand, from paper.
Terra is an AI-native operating system for farms. The Brain connects the operation's fragmented data into one layer and acts on it within the grower's control. The Eyes are computer-vision cameras that capture what is physically happening, acre by acre. Together they are detective and preventative at once: they catch the problem the day it happens and act before it becomes a cost, a violation, or a lost harvest.
02
The pain point
Globally, about 40 percent of crop production is lost to pests and disease every year, more than $220 billion in value. Pesticide use meant to fight that loss carries its own bill: roughly $10 billion a year in environmental and health damage in the United States alone. Both numbers share a root cause. Decisions are made without current, complete information about the field.
The information problem has two halves. First, the data a farm already produces is disconnected. The PG&E bill does not know about the irrigation schedule. The spray log does not know about the weather record. The payroll system does not know who was in which block. No single view of the operation exists, so expensive errors hide in the seams between systems.
Second, the most important data is never captured. A tractor pass sees every tree in the row; almost none of that observation survives the day. Pest pressure, water stress, missing fruit, a stuck valve: by the time these are obvious enough to notice from a truck window, the money is already gone. Margins leave no room for that lag. Input costs have climbed while commodity prices have not, and water in much of California is now metered and capped.
03
The regulatory burden
California grows more food than any other state and regulates farming more heavily than any other state. A single Central Valley operation answers to the county agricultural commissioner, the regional water board, its groundwater sustainability agency, Cal/OSHA, and the FDA, each with its own reports, deadlines, and penalties.
Groundwater: SGMA
The Sustainable Groundwater Management Act requires overdrafted basins to reach sustainability. In practice that means metered wells, pumping allocations, per-acre-foot fees, and extraction reports filed with the local groundwater agency. Basins that fall short face state intervention and steeper fees. A grower who cannot show when and how much each well pumped is exposed on every one of those fronts.
Pesticides: full use reporting
California requires every agricultural pesticide application to be reported to the county agricultural commissioner monthly: product, rate, acreage, location, date, and applicator. Restricted materials require a permit and a notice of intent before the application happens. Each missed or inaccurate report is a separate violation, and violations carry per-incident fines.
Nitrogen: the Irrigated Lands program
Growers in the Central Valley file certified nitrogen management plans and annual summary reports through their water coalition. The numbers have to reconcile: fertilizer purchased, applied, and removed with the crop. Today those numbers are reconstructed from receipts and memory at the deadline.
Labor: overtime and heat rules
Agricultural overtime now begins at 8 hours a day in California, and Cal/OSHA requires a written heat illness prevention plan with training records, shade, water, and rest logs. Hour tracking that lives on paper time cards becomes a liability the moment an inspector or a lawyer asks for it.
Food safety: FSMA
The federal Produce Safety Rule adds water testing, soil amendment records, worker training logs, and traceability requirements, enforced through inspections. Buyers increasingly demand the same documentation on top of the law.
The process is as costly as the rules. Each report draws on records kept in a different place, on a different calendar, by a different person. Compliance becomes an end-of-month reconstruction: pulling receipts, reading meters, calling the applicator, and hoping the numbers line up. The work is cumbersome precisely because it is detective work performed by hand, after the fact, under a deadline.
Here is the opening: every one of those reports is built from data the farm already generates. The burden is not the farming. It is the capture and connection of the data, and that is a software problem.
04
Case study: PG&E demand charges
Agricultural electric rates include a demand charge: a fee set not by how much energy the farm used, but by its single highest draw during the billing month. The rate also varies by time of day, with a peak window in the late afternoon and evening priced far above the rest.
The failure mode is simple and brutal. A well pump is started a few minutes early, or a foreman runs it during the peak window because that is when he got to it. Those minutes set the maximum-demand line for the entire month. The farm pays $3,000 or more for a month in which the same water, pumped a few hours later, would have cost around $200. Nothing on the farm looks wrong. The error is invisible until the bill arrives weeks later, and by then the next billing cycle may already be compromised too. Repeated across an irrigation season, one bad habit on one meter is a five-figure loss, and most farms run many meters.
Terra closes that loop in three moves. It reads what the farm already has: the utility bill, the pump schedule, and the logs the operation keeps anyway. It detects the mistimed start the day it happens by reconciling pump activity against the rate calendar, and flags the billing cycle that is now affected. And it prevents the repeat: it recommends start times that stay out of the peak window and alerts before the next window opens, so the penalty rate never locks in again.
Demand charges are one example of a wider class. Water allocations, spray windows, overtime thresholds, and report deadlines all share the same shape: a timing rule, a record that exists somewhere on the farm, and a cost that lands only when the two are never compared. A system that compares them continuously turns every one of these from a recurring penalty into a solved problem.
05
Detection and prevention
Asked whether Terra is a detective tool or a preventative one, the answer is both, and the order matters. Detection without prevention is a report card on money already lost. Prevention without detection is a guess. Terra runs both on the same data.
Detection is the Eyes and the Brain working together. Cameras capture ground truth from every acre: what is growing, what is stressed, what is missing. The Brain reads every bill, log, and meter and reconciles them against each other. The result is that problems are caught the day they happen, with a record attached.
Prevention is what that record makes possible. The same system surfaces the report before the deadline, flags the application that would break a rule before it is sprayed, and times the pump so the penalty rate never applies. And because every action is captured as it happens, the documentation regulators demand becomes a byproduct of operating rather than a month-end reconstruction. The audit trail writes itself.
06
Where Terra stands
Terra is built in Fresno for the largest agricultural county in America, and it sells the way farming actually works: grower to grower.