A sugarcane field in Panama may not seem relevant to a corn farmer in Iowa.
But Greenwing AI is betting that local crop conditions — especially in regions that are often under-monitored — can provide early clues about broader agricultural risk around the world.
The startup is not trying to tell a farmer whether to spray, irrigate or replant a specific field. It is trying to build an intelligence layer that helps agriculture understand whether local crop stress could signal larger supply pressure, weather risk or market-relevant change.
“The industry is awash in data, but short on context,” says Joshua Fisch, co-founder and CTO of Greenwing AI. “Greenwing is built to cut through the noise and identify material change early and accurately.”
From Satellites to Crop Signals
Fisch came to agriculture from astrophysics, not agronomy. In 2023, he was working at MIT Lincoln Laboratory, applying orbital mechanics and satellite data to national security problems — experience that taught him how to extract useful signals from complex remote-sensing systems.
He soon began asking whether those same skills could apply to agriculture, where satellite data was multiplying but actionable insight remained harder to find.
“There was tons of satellite data coming out from all sorts of different vendors,” Fisch says. “What was missing was, well, how can I act on that? Where’s the actionable insight there?”
That question became the starting point for Greenwing, which Fisch began building during evenings and weekends. Today, the company is focused exclusively on crops — not livestock, forestry, shipping or transportation infrastructure.
“We’re really focused on this very specific question of agronomic intelligence and the health side of it,” he says.
The First Alert: Sugarcane in Central America
Greenwing’s first Agricultural Intelligence Alert, released July 21, shows how the company wants its system to work.
The alert focused on sugarcane in Panama and Costa Rica, where Greenwing identified “a growing soil-moisture deficit across Central American sugar-producing zones, consistent with the onset of El Niño conditions.”
The report carried Greenwing’s highest confidence rating, Level 4 of 4. It covered sugarcane-producing areas totaling 167,197 hectares in Panama and 97,088 hectares in Costa Rica, with observations refreshing within 120 hours at 100-meter resolution.
For U.S. corn and soybean farmers, Central American sugarcane may sound distant. But the alert offers a concrete example of Greenwing’s larger goal: identifying crop stress signals early, before they fully show up in the plant canopy or in traditional production estimates.
In Panama, Greenwing found soil moisture deficits had persisted through three consecutive observation cycles along the Pacific dry margin. Yet canopy vigor remained broadly favorable, meaning the deficit had “not yet surfaced as visible crop stress,” except for early thinning in one cane basin.
That matters because Greenwing is not simply flagging crops that already look bad from space. It is trying to interpret when an underlying condition — such as moisture stress during a water-sensitive growth stage — may become agronomically significant.
“These deficits occur during a water-sensitive stage,” the alert stated. “Greenwing’s nine-year prior seasonal studies show that deficits at this stage have preceded tighter regional crop supply.”
Why Context Matters
Greenwing uses satellite data, including European Space Agency Sentinel satellites refreshed as often as every five days, or within 120 hours, at 100-meter resolution.
Fisch says the value is in interpretation.
“Our models use current data and up to nine years of history,” Fisch says. “We interpret each signal within its regional, seasonal, and crop-specific context to identify meaningful change.”
In practice, that means Greenwing measures each zone against its own multi-year, calendar-matched norm. The company is not only asking whether a region looks dry. It is asking whether that region looks meaningfully different from what would be expected for that crop, in that place, at that time of year.
“I would say we are essentially in the business of interpretation,” Fisch says. “What Greenwing provides to this whole problem is context and interpretation. You can interpret the same data in millions of different ways, but people only care if you interpret it properly.”
Why Local Crop Signals Could Matter Globally
The company describes its alerts as an intelligence layer for people making decisions tied to agricultural production.
“Our Agricultural Intelligence Alerts provide growers, agribusinesses, investors and governments with timely, reliable intelligence on material changes in agricultural conditions,” Fisch says.
This kind of system could help growers and agribusinesses understand whether crop stress is isolated, regional or part of a larger global pattern.
Central America is a marginal contributor to global sugarcane production, but Greenwing says that is precisely why the region can be useful during broad weather events. In its alert, the company stated: “Its significance rests in its role as a Greenwing-established leading indicator in specific global events.”
In this case, the broader event is the projected 2026-27 El Niño. Greenwing says its historical analysis shows drying along the Central American Pacific margin can serve as an early marker of wider agronomic trends consistent with lower global sugar supply.
Fisch says the same logic could eventually matter for U.S. row crops.
“The notion that U.S. corn and soy exists in its own isolated chamber and bubble is becoming further and further from the truth,” he says.
What Greenwing Says It Does — and Doesn’t — Do
Fisch says the company is not predicting prices, recommending trades or forecasting yields.
“We very intentionally do not predict prices or yields,” he says. “We are strictly monitoring and reporting on a sort of neutral intelligence basis.”
For now, Greenwing does not have paying customers, by design. Fisch says the company is making its first signals and reports public before moving into paid products. Potential customers include agribusinesses, governments, insurers, commodity traders, other intelligence firms and growers.
The Gro Intelligence Question
Greenwing is entering a category where agriculture has seen both big ambition and big disappointment.
Any ag data startup working in global crop intelligence will invite comparison to Gro Intelligence, the once-prominent ag analytics company whose collapse left many in the industry skeptical of big promises around agricultural data.
Fisch says Greenwing is working on a related problem — making agronomic data intelligible and actionable — but approaching it differently.
“How can we make all this awesome agronomic data that’s flowing throughout the world intelligible and actionable?” he says. “There’s sort of the ground-up approach, which is that you have a team of agronomists that are learning about signal intelligence, and signal-to-noise ratio, and all of these different awesome technical things. And then you have sort of the top-down approach that we’re building here, which is a very physics-based, satellite-based, sensor-based approach.”
Fisch says Greenwing’s team is intentionally interdisciplinary, combining satellite and data expertise with agricultural context. The company works with 16 PhD-level scientific advisers, including data scientists, agronomists and farmers, with an emphasis on regional and geographic diversity.
“It’s always going to be interdisciplinary,” he says. “There’s never one person or one unit that is going to be able to solve this problem.”
Fewer Reports, Stronger Signals
Greenwing also says it is prioritizing quality over volume. Rather than issuing reports on a fixed schedule, the company plans to publish alerts only when its models identify a signal the team believes is meaningful.
“If a signal happens three times a week, we’ll get it out,” Fisch says. “If it doesn’t happen for a month, we’ll wait a month. We don’t feel any pressure to ever communicate anything to the public that we wouldn’t 100% stand behind.”
The company says it currently monitors 16.9 million hectares across 38 countries and plans to expand.
“We’ll continue publishing selected Agricultural Intelligence Alerts as events warrant, while expanding beyond the 16.9 million hectares we currently monitor across 38 countries,” Fisch says.
Asked what success looks like one year from now, Fisch says the goal is trust.
“We’re hoping to continue to be a trusted source of agronomic intelligence that people can rely on,” he says. “We want to continue to be a quantitatively and logically driven platform that people can rely on and trust for early and precise agronomic warnings and intelligence.”
