Kris Covey

Kris Covey is building the first public data infrastructure for soil—the foundation for a transition that could turn agriculture from climate liability into climate investment. His insight: regenerative practices will not scale until farmers, markets, and lenders share measurement systems they can trust.

Kris Covey’s core insight is that agricultural data must function as public infrastructure if the world is to transition to healthy, nutrient-rich soil for sustainable agriculture at scale. Today, that transition is stalled by a profound trust deficit: farmers, markets, and investors cannot reliably measure or verify the outcomes of soil stewardship. Without credible, transparent, and accessible data, regenerative practices remain underfunded and misunderstood – locking in ecological degradation, food insecurity, and economic instability for hundreds of millions of producers.

This measurement gap is systemic. Markets cannot reward practices they cannot verify; governments cannot design effective incentives without reliable baselines; and farmers cannot access finance or participate in emerging carbon markets without trusted evidence of their impact. Agriculture, responsible for a major share of greenhouse gas emissions and biodiversity loss, lacks the shared measurement systems that have enabled progress in forestry and meteorology. The result is a field facing both urgent risk and extraordinary potential.

Through The Soil Inventory Project (TSIP), Kris is building the first public data infrastructure for soil, and in so doing, he’s making agricultural measurement trustworthy, transparent, and universally accessible. TSIP combines advances at three levels into a coherent, field-shaping system. Technically, the team has built what was missing: low-cost, durable sampling hardware; a software platform (Fieldvision) that automates sampling design and analysis; and, launched in 2025, the first multi-model ensemble for agriculture – an eight-model prediction system that improves accuracy by more than 40 percent and brings the rigor of weather forecasting to soil carbon. Institutionally, TSIP’s nonprofit structure enables binding commitments to data transparency and protections that no private actor can credibly promise, allowing it to hold soil data in public trust. Socially, TSIP has created a reciprocal data system in which contributors receive immediate, high-value insights, treating equity as a prerequisite for scientific validity.

By building infrastructure designed for absorption into public systems, TSIP is laying the foundation that public agencies, markets, and producers will ultimately rely on. As government capacity catches up, TSIP’s standards, tools, and datasets can be absorbed into public programs. In the meantime, Kris is occupying the ground neither government nor private industry can hold: technically sophisticated, institutionally trustworthy, and oriented entirely toward the public good, unlocking the credible measurement on which regenerative agriculture depends.

The forest carbon market is a multi-billion-dollar industry. When consumers offset a flight or a purchase, most of that money flows to someone doing something with a tree. This market exists because we know how to count trees. A century of federal investment through the U.S. Forest Service produced standardized methods: identify the species, measure the trunk, look up the carbon content in a table.

The annual flux of carbon into agricultural soils dwarfs what trees can capture. Yet the soil carbon market remains a fraction of the forest market's size. The reason is straightforward: we cannot count soil the way we count trees. There is no equivalent table. There is no simple measurement that tells you, with confidence, what you are looking at.

Soil is the farm’s built-in buffer. It determines whether rain soaks in or runs off, whether fields hold moisture through drought, how steadily crops can be fed without constantly escalating inputs. USDA findings show soil-related problems are already common across U.S. cropland, meaning a significant share of production is operating with less resilience than it appears. If current practices continue, the likely outcome is not a single dramatic collapse but a steady erosion of that buffer: more frequent bad years, rising input costs, greater vulnerability to the droughts, floods, and temperature swings that climate change is already delivering. In grain-producing regions, mid-century projections point to meaningful yield pressure for major crops. In areas dependent on irrigation, aquifer drawdown is forcing a hard choice: pay more to chase deeper water, or shift acres out of production and accept lower, riskier returns.

Regenerative practices, cover crops, reduced tillage, diverse rotations, integrated grazing, can rebuild what conventional systems wear down. The benefits cascade: carbon sequestration, reduced fertilizer dependence, cleaner waterways, improved yields over time, greater resilience to drought and flood. Whether you care about climate, water quality, farmer livelihoods, or food security, regenerative agriculture offers a clear path. The practices are well-documented. The science is not in dispute.

Yet of 900 million arable acres in the United States, only about 1.5 percent is farmed regeneratively. Adoption stalls on a simpler problem: nobody can put reliable numbers on the benefits. The situation is like telling someone to save for retirement without disclosing their salary, the return on their portfolio, or even the cost of checking their bank balance. You cannot tell them how much more they will earn if they work harder. You cannot tell them what their new practice will yield. But you insist they should really do it. Then someone else arrives the next day offering a Porsche, right now, no uncertainty. That is essentially what the fertilizer companies can do. They specify exactly how much nitrogen to apply, with which variety, in which county, planted on which date—and they predict the yield within narrow margins. Loan financing is built around those projections. At present, regenerative agriculture offers stories where conventional farming offers spreadsheets.

The harder problem is that measurement produces confusion rather than clarity. Soil carbon assessment relies on biogeochemical models—mathematical representations of how carbon moves through agricultural systems in response to climate, weather, and management. These models are trained on different datasets, optimized for different conditions, built with different assumptions about which biological processes matter most. They routinely contradict one another. A farmer or investor trying to project the benefits of a practice change encounters several answers, often diverging by wide margins. That variability is fatal to confidence. Organizations want to reward farmers for sequestering carbon, but without trusted projections, the risk of adopting new practices is too high for most producers and investors to bear.

The measurement gap falls hardest on small landholders. Sampling costs are prohibitive, the technical complexity is high, and the distrust is earned: small-scale producers have watched others profit from their data while receiving nothing in return. Many fear that sharing information will benefit corporations, not them—or that verified carbon data will eventually be used to mandate practices rather than reward them. But models trained only on large conventional operations cannot represent the full picture. Smaller farms, with their varied practices, different soils, and experimental spirit, supply precisely what the system needs to be valid. Without that data, the models stay miscalibrated, and the feedback loop that could accelerate adoption never forms.

Attempts to solve this have failed repeatedly. Federal programs cannot operate at the necessary scale or speed. Universities produce valuable research but lack infrastructure to deploy it nationally. Private companies have harvested farmer data to sell products, not to serve producers. John Deere, Cargill, and their peers hold the largest data pools in agriculture—gathered as a byproduct of the equipment farmers already run and the inputs they already buy, returned to farmers at no apparent charge. The actual transaction is the data itself, flowing back to entities whose business is selling inputs. A measurement system built to sell nitrogen cannot function as neutral evidence of carbon outcomes; no audit transforms a proprietary single model with closed methodology and no published uncertainty into something markets, auditors, or regulators can rely on. For-profit verification firms face an even more corrosive problem: they are paid by the claims they validate. This creates pressure to deliver favorable numbers, not accurate ones. The result is a patchwork of competing tools, incompatible datasets, and eroded trust—precisely the opposite of what a functioning market requires.

Weather forecasting solved an analogous problem decades ago. Rather than selecting a single atmospheric model and hoping it proves correct, meteorologists run multiple independent models and layer their outputs into ensemble systems. When predicting hurricane paths, ensemble methods outperform any individual model precisely because they acknowledge uncertainty while producing more reliable forecasts. No one would trust a hurricane prediction built on a single model. Yet that is exactly what agriculture has been doing with soil—picking one model and hoping it is right, or worse, selecting whichever model produces the preferred result.

The infrastructure to do for soil what meteorology did for weather has not existed...until now.

TSIP’s strategy begins from a single premise: agricultural data should function as public infrastructure. Every major advance in land management has depended on shared measurement systems. The forest carbon market exists because a century of federal investment produced standardized methods for counting trees. Weather forecasting works because meteorologists pool data and run multiple models in ensemble. Soil has neither: no shared inventory, no ensemble modeling, no neutral infrastructure. TSIP is building that infrastructure now.

Since 2019, the organization has worked across three reinforcing dimensions: technical systems that make measurement possible, institutional commitments that make it trustworthy, and social architecture that drives adoption without requiring ongoing subsidy.

The technical work began with the tools themselves. Kris developed a low-cost drill-auger field kit that allows any producer to collect accurate soil samples two to three times faster than traditional methods. The kit works where standard push-probes fail, in dry, compacted, or rocky ground, and costs a fraction of commercial sampling services. A paired phone app guides users through the process: outlining fields, navigating to GPS-marked points, syncing barcoded sample bags to precise locations. Geolocated samples are sent to be processed at commercial laboratories and reflected as automated reports within days, complete with regional comparisons. For smaller farmers who have been priced out of private services and underserved by infrequent state sampling programs, this is access that did not previously exist.

Better sampling alone cannot overcome the deeper problem of conflicting models. TSIP's central technical contribution is the first agricultural multi-model ensemble, developed in collaboration with Michigan State and published in Nature Scientific Reports in 2025. The ensemble integrates eight peer-reviewed biogeochemical models and delivers a consensus prediction with quantified uncertainty. It outperforms any single model by 40%. It also measures both carbon sequestration and nitrous oxide emissions, and it compares regenerative practices against a dynamic baseline (what would happen under continued conventional management, not a static snapshot). This structure allows markets to reward verified climate impact. The tool is currently available for the Midwest, with other regions to follow.

The institutional structure is equally deliberate. TSIP operates as a nonprofit, forgoing access to venture capital and other funding streams that a for-profit model would enable. This constraint is the point. In a field where proprietary models and opaque data ownership are competitive advantages, TSIP can make commitments no for-profit actor can credibly offer: it will always publish results and it will act solely in the public interest. The structure allows TSIP to hold sensitive information in trust, serve farmers who cannot pay, and follow the science even when findings are inconvenient. Federal agencies lack the technical infrastructure to steward national soil data; farmers (especially those historically harmed by government programs) have reason to distrust federal involvement. TSIP occupies the middle ground between them: technically sophisticated, institutionally neutral, and accountable to no one’s bottom line.

At $2.5 million against a multi-billion-dollar industry, competing on data volume is a race already lost. TSIP’s answer runs on two tracks. The first is technical: where commercial actors deploy a proprietary single model with no published uncertainty, TSIP’s multi-model ensemble produces peer-reviewed, publicly auditable consensus predictions, published in Nature Scientific Reports. When a carbon registry, a lender, or a federal baseline program must choose between a commercial black box and a documented ensemble, data volume does not close that gap. The second track is institutional: because TSIP holds data in public trust and publishes its methods, its measurements can travel into contexts where single black-box models cannot go.

A farmer can use John Deere’s precision agriculture tools to optimize inputs and TSIP’s infrastructure to certify carbon outcomes for a market payment or supply chain requirement. These serve different functions. TSIP does not need to displace the giants; it needs to become the standard above them—the layer against which any measurement, including theirs, is ultimately judged. That position is defensible because the function it performs cannot be performed by an entity with a commercial stake in the result.

What induces farmers to choose practices whose environmental benefits exceed their immediate economic return? The answer is market design. Regenerative practices are financially irrational for most producers today because their benefits are difficult to verify credibly, and therefore chronically undercompensated. Credible, independent measurement changes that calculus: it is what allows a carbon market to pay a farmer for sequestration, a lender to underwrite a practice transition, a supply chain buyer to meet sustainability commitments with auditable evidence. Supply chain requirements are already functioning as a quiet mandate—when a cooperative commits to verified sustainability, its suppliers face a business case no public campaign could manufacture. TSIP is building the system that makes environmental and economic interests converge.

Driving adoption is the third dimension. TSIP’s data system runs on reciprocity: contributors receive immediate value in the form of reports and projections that would otherwise cost thousands of dollars. Each new dataset improves model accuracy, which increases the value of participation, creating a feedback loop. “Data equity is data integrity” is a design principle—smallholders access the same tools as corporations because models calibrated only to large conventional operations cannot represent the full picture.

Technical rigor without adoption is a research project, not a system change. Rather than recruiting farmers directly, TSIP works through trusted intermediaries who carry the tools into their own communities. Nature for Justice, a network of more than 175 Black farmers in North Carolina and Virginia, brings TSIP’s sampling infrastructure to producers who have reason to distrust outside institutions. Stonyfield connects TSIP to its network of organic dairy suppliers. Bayer provides data at industrial scale. The Foundation for Food and Agriculture Research—created by Congress to connect agricultural science with producers, and working directly with commodity groups and farm organizations nationwide—has partnered with TSIP. That endorsement reaches farmer networks that no startup, nonprofit or otherwise, could access alone. TSIP provides infrastructure; trusted partners provide reach and engagement.

One test of this model came through the USDA's Climate-Smart Commodities program. In September 2022, TSIP was awarded approximately $20 million to lead a five-year project spanning 120,000 acres. Partners included Nature for Justice, Jackson Family Wines, and the Glynwood Center for Regional Food and Farming. The project enrolled farmers growing everything from Midwest row crops to specialty wine grapes, with more than a quarter of funding directed to small and underserved producers. TSIP's role was measurement, monitoring, reporting, and verification of soil carbon across the portfolio.

In April 2025, the Trump administration cancelled the program, a devastating blow to the organizations involved and the foundational work of the first two years. Yet farmers who had been paid to use TSIP’s tools kept using them without compensation. Once farmers had the infrastructure in hand, they did not need to be paid to use it. They needed only to see that it worked.

TSIP now operates with a budget of $2.5 million (up from $400K in 2024) and is building toward a revenue model in which large corporations pay for platform access while smallholders use the tools at no cost. The companies that benefit most from verified sustainability claims fund the infrastructure that makes verification possible for everyone.

The long-term aim is to become for agricultural soil what the National Weather Service is for atmosphere and what the Forest Service built for trees. If TSIP succeeds, the 1.5 percent of farmland now managed regeneratively becomes a floor rather than a ceiling. Farmers had the evidence to act in their own interest. That was enough.

Kris was raised primarily by his mother, a former nun who went on to earn a PhD while managing multiple sclerosis. Often left to make his own fun, Kris spent his childhood in the woods near his house—designing forts, digging underground caverns, losing himself in projects. During summers, his father would take him on months-long camping trips through the American West, sleeping under the stars in national parks.

After college, Kris spent years as a professional whitewater kayaker, testing boats for Dagger Canoe and Kayak. Eventually he took a job teaching middle and high school science in rural Massachusetts. One of his classes, by quirk of state curriculum, attracted both college-bound seniors seeking a fourth science credit and students who had failed chemistry. The valedictorian sat next to a student who struggled to read at a high school level.

Traditional methods weren’t reaching everyone, so Kris built something different. He connected with local arborists who were struggling to find labor and pitched the school board on creating a campus arboretum. Some students ran GIS systems and produced landscape planning documents. Others nurtured saplings, dug holes, and learned to identify plants. The tree companies got a workforce pipeline; the community got green space; the students got skills and shared purpose. What looked like a teaching problem turned out to be an infrastructure problem. He discovered he was good at building it.

That instinct carried him to Yale School of Forestry and Environmental Studies, where he spent a decade as a master’s student, then a PhD candidate, then as a Postdoc, Lead Scientist for a Research Center, and Lecturer in Forest Dynamics. In addition to research, Kris ran orientation programs for incoming forestry students. When he noticed that disciplines weren’t intermingling, he raised funds from the Dean to install a commercial espresso machine in the science building, drawing policy students into conversation with researchers. One morning over coffee, a colleague brought a graph estimating how many trees exist on Earth. Kris initially dismissed it—all they had discovered, he argued, was that big trees take up more space than little trees. But the question stuck. The resulting collaboration, with contributors from around the globe, produced a reliable estimate that Earth holds three trillion trees and landed on the cover of Nature.

The tree-counting project revealed to Kris that large-scale inventory work could produce consequential science. Working with ranchers in Wyoming, Kris asked what kind of data could change their operations. The answer was soil carbon. When he turned his attention to soil, he found a field in chaos: conflicting models, proprietary data, no shared infrastructure. That gap led to QuickCarbon, an early research program exploring rapid soil carbon assessment, and eventually to TSIP.

As of fall 2025, Kris is focused full-time on TSIP. His former students work for him. Collaborators describe him as a natural convener, someone who builds trust between unlikely groups and channels it into collective action. The arboretum still stands in Massachusetts. The espresso machine still runs at Yale. TSIP is the same instinct at different scale, building the permanent infrastructure that makes the change possible.