I started this publication to say plainly what the industry usually only says in hallways. I did not know how many of you had been waiting for someone to do that. Now I do. It has meant a lot.

I have sat in three chairs.

Seven years as a grower, Avocadoes, cotton and Flowers, where you learn what a recommendation costs when it does not work and you are the one who made the decision. Twenty years inside the input industry — formulation, agronomy, regulatory, marketing, and selling — across plant nutrition, crop protection, adjuvants, and precision agriculture. Five years as a consultant auditing that system from the outside, where you learn that the view from any one chair is radically incomplete when taken alone

.

Most people in agriculture have lived one of those chapters. The arc across all three is what this publication has been built on.

This is the final Field Note in this series. I want to use it differently than the rest. Not to diagnose another failure point in the system, but to describe the farm I would actually build if I were starting over as a grower today.

This is the farm I’d build today.

1. Build the advisory team around the acre

The first decision is the advisory relationship, and it would be the most different from what most growers currently have.

I would not have a dealer agronomist as my only advisor.

This is not a new idea. The best growers figured this out a long time ago. Walk any progressive operation and you will find an independent agronomist somewhere in the picture.

That does not mean the dealer agronomist has no place on the team. They do.

A good dealer agronomist keeps you current on what is moving through the pipeline — new formulations, label changes, emerging resistance patterns, products worth watching. That intelligence has real value.

The problem is not the dealer agronomist.

The problem is making them the primary source of integration when the system they operate inside rewards them for something other than integration.

I would have an independent agronomic consultant whose compensation is tied, at least in part, to documented outcomes on my operation: margin, consistency, avoided mistakes, better placement, and the renewal discipline of the program.

Not tied to what I buy.

Tied to what the buying returns.

That single shift in accountability — not the credential, not the years of experience — is what changes the advisory relationship from a channel I am the end of to a partnership I am the center of.

There is a third seat at this table that most growers never formally assign: the person or institution financing the operation.

Whether that is a bank, a family arrangement, or outside investors, they have a stake in every agronomic decision made on the farm. In my experience, the ones who are kept informed and treated as part of the team make better partners than the ones who only hear from you when something goes wrong.

Some lenders and investors have their own agronomists watching the operation. I would want those people in the room, not at arm’s length. Keeping them current on what I am doing and why — the evidence behind the program, the logic of the advisory team, the direction of the operation — builds the kind of trust that creates flexibility when you need it most.

This is the role I described in Field Note 11 as the original agronomist function: the generalist whose job is integration across every category, accountable only to the acre.

I would pay for that function directly, the way I pay for any other professional service that is genuinely accountable for its results. And I would evaluate it the same way — on what it returns, not on whether I like the person delivering it.

2. Raise the evidence standard

The second decision is the evidence standard I would hold myself and my program to.

I would not accept a product recommendation without the conditions.

Not the trial result. The trial conditions.

The hybrid the data was generated on. The soil type. The application timing relative to growth stage and weather. The input program the product was tested against. The stress profile of the season. The fertility background. The crop protection history. The operational context around the pass.

If the conditions do not travel with the number, the number does not tell me anything useful about my ground, and I would not pay to find out the hard way.

This sounds like a higher bar than the industry currently sets.

It is.

And it is achievable — not because I would wait for manufacturers to improve their data packages, but because I would generate my own evidence.

Structured strips. In-season tissue and sap analytics. Remote sensing calibrated against what I am actually trying to measure. Field observations logged against timing, growth stage, weather, and application conditions. On-farm data that gets analyzed rather than accumulated.

The data matters. But understanding how a technology actually works — its mechanism of action, what it is doing at the biological level and why — is equally important, and sometimes more so.

If I understand the mechanism, I can reason about where it fits in my system: which conditions activate it, what might suppress it, how it interacts with the rest of the program. That understanding is what lets me place a product precisely rather than apply it hopefully.

It also tells me what I should be measuring.

Because if I know what the technology is designed to do at the biological level, I can build the right detection method around it rather than reaching for whatever measurement is easiest or most available.

There is also a category of tools I would evaluate differently than most: the ones designed not to produce dramatic results every season, but to perform reliably across eight out of ten.

I think of these as agronomic insurance tools.

They may not show up in a single-year strip trial the way a high-response product does. But on a farm managed across a rotation, under variable weather, with real margin pressure, consistency has economic value that a peak-performance measurement misses entirely.

I would keep a short list of those tools and not apologize for it.

The grower as the auditor of last resort — which is what “The Input Stack, Audited” named — is not a sustainable workaround. It is an institutional failure that I would compensate for personally, on my own operation, with my own tools, until the system catches up.

The knowledge I generate would not stop at the fence line.

My consultant would see it. The companies whose products I am evaluating would see it, in the form of renewal decisions explicitly tied to what I could or could not document.

No product recommendation would move forward because the story was good. It would move forward because the evidence was good enough for the acre where the check is written.

3. Design the input stack before buying the pieces

The third decision is how I would build the input program itself.

I would start with the stack, not the category.

Every input I consider would be evaluated not just for what it does in isolation, but for what it does inside the program I already have — and what it requires the program to become in order to work properly.

A biological inoculant that needs a different starter fertilizer to perform well is not a conversation that starts after I have already bought the inoculant.

It is the first conversation.

A biostimulant that changes plant signaling may change the fungicide decision later in the season. A fertility product that alters early growth may change water demand, tissue balance, or disease susceptibility. A biological that performs under one residue-management system may behave differently under another.

These are not product questions alone.

They are program questions.

This means I would work with my team — not a product representative — to design the program as a system before committing to its components.

The trial data is a starting point, not a decision.

The question is always whether the product’s characterized performance holds under my conditions, inside my stack, with my operational constraints.

For biologicals specifically, I would apply a visibility standard alongside the efficacy standard.

If a biostimulant is doing something real at the physiological level, we should be able to define where we expect the signal to show up — tissue chemistry, root architecture, stress response, nutrient-use efficiency, growth rate, fruit quality, stand resilience, or yield stability — even if the signal is conditional and not always easy to capture.

That matters.

A biological does not have to produce a dramatic visible response every season to be valuable. But I do need to know what I am looking for, when I should expect to see it, and under which conditions the response should become measurable.

If we cannot design a detection method alongside the application, I would not commit the product to the program.

Not because I distrust the science.

Because I cannot manage what I cannot see, and I cannot justify what I cannot document.

4. Use crop intelligence as operating infrastructure

The fourth decision is the technology layer, and this is where the farm I would build today is structurally different from the one I farmed thirty years ago.

I would use crop intelligence as operating infrastructure.

Not as an add-on after the agronomic decisions are made, but as the synthesis layer that makes integrated agronomic decisions possible in the first place.

What my team recommends would be informed by a system that holds the full picture simultaneously: microbiome conditions, nutrient removal budget, spray schedule, weather forecast, residual chemistry, hybrid or variety, irrigation status, field history, labor constraints, equipment windows, and the operational reality of my specific farm.

This is not AI replacing the agronomist.

It is the exogram that talks back — an external memory system that can hold the pieces of the operation together when complexity exceeds what any one person can carry alone.

For the full argument on exograms and agriculture’s long lineage of external memory tools, see Field Note 3: “The Agronomist and the Algorithm,” and Bentz & Dutkiewicz, PNAS, 2026, DOI: 10.1073/pnas.2520385123.

The point is not that software suddenly knows more than the agronomist.

It does not.

The point is that the farm now generates more interacting information than any unaided advisor can hold in a single frame. The role of crop intelligence is to preserve the relationships between those pieces long enough for better decisions to be made.

The same expertise.

Full radar.

The recommendation I receive would be calibrated to my ground, my genetics, my conditions — not extracted from a guide built for a composite of operations that may or may not resemble mine.

The difference between those two is the difference between a product that performs in the trial and a product that performs on my acre.

That difference is where the margin lives.

In practice, this farm would run on a different operating rhythm.

Pre-season, the advisory team would build the program from the stack up, with every major input tied to a hypothesis, a placement condition, and a measurement plan.

In-season, the program would be adjusted against tissue, sap, imagery, weather, pest pressure, crop stage, field observations, and the reality of the week’s workload — not calendar habit alone.

Post-season, every major decision would be audited for return, fit, and repeatability.

Products that earned renewal would stay.

Products that only earned explanation would not.

5. Enter the development process earlier

The fifth decision is the posture I would bring to the system as a grower.

I would make it known — to the companies whose products I use, to the researchers whose work I follow, to the developers whose pipelines I watch — that I am willing to participate in the development journey.

Not every company will take you up on it, and not every development process has room for it.

But some do.

And the grower who raises their hand early, who offers their operation as a learning environment rather than just a sales target, tends to end up with better products, better timing, and better relationships than the one who waits for the finished label to arrive.

Field Note 1 described a system where the grower enters last and inherits a program they did not shape.

I would try, wherever possible, to enter earlier.

That does not mean turning the farm into someone else’s R&D department. It means being deliberate about the relationship. Clear expectations. Clear plots. Clear measurements. Clear ownership of what gets learned. Clear renewal decisions when the season is over.

And when I learn something real about how a product or a practice performs on my ground, that learning would not stay on my operation.

It would flow back through the advisory relationship, into the evidence base, toward the next grower making the same decision.

The horizontal network of peer-to-peer knowledge that growers have always relied on — which Field Note 6 named as the most trusted information source in the system — would be intentional, not incidental.

Not coffee-shop rumor. Not plot-tour theater. A disciplined, grower-centered evidence network. That is how the acre starts talking back to the system.

What this farm actually replaces

The farm I would build today is not hypothetical.

The components exist. The tools are deployable. The advisory model already has proof of concept in crop systems where independent consultants have operated on accountability-based relationships for a generation.

What is still being built is the institutional infrastructure: the incentive alignment, the evidence architecture, the technology integration, and the renewal discipline that make this farm the norm rather than the exception.

That infrastructure is being built now.

By the growers who have already crossed the complexity threshold. By the consultants who recognized that the only sustainable practice answers to the acre. By the companies that decided the evidence problem was theirs to solve. By the operators who understand that margin is no longer protected by doing what worked ten years ago and hoping the system still holds.

I would build this farm because the arithmetic supports it: fewer acres exposed to unproven recommendations, fewer products renewed on habit, better placement of tools that do work, fewer avoidable passes, fewer surprises hidden inside the stack, and a clearer record of what actually returned money.

Not because the system has made it easy.

It has not.

But the system designed for a simpler century is no longer adequate for the agriculture we are actually practicing.

The reckoning is not an event.

It is the gradual replacement of a design that no longer works by one that does.

The farm I’d build today is part of that replacement.

And it starts with growers willing to build the discipline before the system makes it easy.

One more thing before I close out this series: I want to hear from you.

The next set of essays is still taking shape, and I would rather build it around the questions you are actually sitting with than the ones I assume matter.

What are you seeing in your operation, your market, or your advisory relationships that the industry is not talking about honestly?

What would you want this publication to dig into next — a specific input category, a business model question, a technology you are trying to evaluate, or a failure you have watched play out in slow motion?

I read every reply.

Hit respond and tell me what is on your mind.

That is the conversation this was always meant to start.

— Vatren Jurin, Partner, DunhamTrimmer

The Reckoning publishes monthly Anchors and weekly Field Notes — free, always.

https://www.quantamagazine.org/the-hidden-mathematical-dance-inside-plant-cells-20260504/

Then the input stack specialized. Fertility became its own discipline. Crop protection became its own discipline. Seed genetics, soil science, entomology — each carved into a domain with its own credentials, its own literature, its own career track. The specialists went deeper. The generalist function — the one that held the whole system in view — quietly disappeared from the institutional structure. It did not become unnecessary. It became uncompensated. And what the system cannot pay for, it stops producing.

The agronomic consultant emerged because the farm still needed that function, even after the institutional structure stopped producing it.

Most experts in the advisory chain answer to an institution outside the farm. The district sales manager answers to volume. The dealer agronomist answers to the portfolio their employer carries. The extension specialist answers to a grant and education mandate. Each of them is skilled, often excellent, and structurally prevented from doing the one thing the farm actually needs: integrating across all of it with accountability only to the outcome.

The agronomic consultant answers to the acre.

That shift in accountability — more than credential, training, or tenure — is what specialization removed and what the independent consultant restores. Not a new role. The original one, operating outside the structure that eliminated it.

Specialty crop agriculture understood this first because it had no choice. Tree fruit, wine grapes, and vegetables have always carried more decision complexity per acre, higher economic exposure per decision, and less tolerance for the integration gaps that row crops could absorb through scale. A walnut grower who missed a critical disease window lost yield on a canopy that took a decade to establish. There was no next-year adjustment. The consequence was immediate, irreversible, and personal. That reality created demand for the agronomic consultant decades before row crop agriculture felt the same pressure.

But the more consequential part of the specialty crop story is what that advisory model positioned those consultants to see first. When biological-based inputs began moving from research pipelines into commercial fields, specialty crop agronomic consultants were among the most capable early evaluators — not because they had unique biological training, but because their systems had already trained them for the kind of management biological inputs demand.

Their performance was shaped by soil microbiome conditions, plant physiological state at application, and interactions with residual chemistry. These complications do require new technical literacy, but they do not require a new advisory logic. They require the integrated, systems-level thinking that specialty crop consultants had already been practicing by necessity for a generation.

They were not just early adopters of biological technologies. They were the professionals who learned, at commercial scale, what biological inputs actually need to perform.

Here is where convergence matters. Biological inputs do not change their requirements based on the cropping system. A microbial inoculant’s dependence on soil microbiome conditions and application timing is not different in a corn field than in a blueberry planting. A biostimulant’s interaction with plant hormonal state does not recognize the distinction between row crops and horticulture. The management complexity that biological technologies carry is consistent across systems — which means the advisory model required to deliver them is also consistent.

The specialist-and-handoff architecture that row crop agriculture normalized was never sufficient. It was tolerable when the inputs were simple enough that integration loss was survivable. Biological inputs are not that simple. Their performance lives in the interactions — between the biology and the soil, between the application and the fertility program, between the timing and the crop’s physiological moment. No category specialist, accountable primarily to their domain, is positioned to manage those interactions completely.

The agronomic consultant is. And specialty crop agriculture spent forty years proving it.

Row crops are arriving at the same inflection point now — driven by the same biological complexity, compressed by the same margin pressure, and inheriting the same advisory gap that high-value horticulture closed decades ago. The difference is that row crop producers are arriving with something specialty crops did not have at the same stage: AI-assisted workflows that can begin to organize the full system at a resolution no individual advisor could previously manage alone. Microbiome profiles, nutrient plans, spray schedules, weather windows, operational constraints — synthesized in service of the integrated judgment the agronomic consultant was always being asked to make.

The technology does not replace the agronomic consultant. It restores leverage to the function the original agronomist was supposed to perform before specialization made the complete job structurally impossible.

The new system did not create the need for the agronomic consultant. It removed the last excuse for not having one.

Vatren Jurin, Partner, DunhamTrimmer

The Reckoning publishes monthly Anchors and weekly Field Notes — free, always. Read the full first Anchor, “The Misaligned Machine,” and now the second Anchor “After the Agronomist”https://vatrenjurin.substack.com/p/the-reckoning

Not too complex for the science. The science is extraordinary. Microbial consortia engineered for rhizosphere colonization. Peptide-based elicitors priming systemic resistance pathways. Plant extracts capable of influencing both pathogen defense and drought response through signaling systems the industry is only beginning to understand. Synthetic growth regulators operating at parts-per-million precision, where efficacy and phytotoxicity may be separated by single-digit concentration changes interacting with temperature, growth stage, and cultivar genetics.

The biology has leapt forward. The recommendation infrastructure — the accumulated layer of agronomic experience, generational field knowledge, advisor judgment, trial data, and operational intuition that determines what goes on which field — has not.

That gap is no longer a quality-of-advice problem. It is a category-survival problem.

Modern agriculture built its advisory systems around chemistry. A synthetic pesticide operates through a relatively defined mechanism against a relatively defined target. One product. One mode of action. One set of conditions that matter. The infrastructure surrounding agronomy — from sales organizations to recommendation guides to replicated trial structures — evolved around this level of complexity.

Biological systems do not behave this way.

A microbial inoculant enters a living ecosystem whose response varies with soil type, moisture, organic matter, crop rotation, residue load, and the microbial legacy of management decisions accumulated over years. A biostimulant influencing root architecture acts through hormone pathways simultaneously shaped by temperature, water status, fertility balance, and cultivar genetics. A biological crop protection organism functions inside a trophic system where persistence and efficacy depend on what else exists in the canopy and soil food web.

The biological is the interaction.

Strip the interaction away and what remains is a label and a set of claims generated under conditions that may not exist on the field where the product will actually be used.

That reality is now colliding with economics. As remarkable as biological technologies have become, many categories are already drifting toward commoditization. When every manufacturer has a Trichoderma platform, when plant extracts become interchangeable SKUs differentiated primarily by price rather than placement precision, the category begins collapsing toward the same margin compression seen in commodity chemistry.

The antidote is not better marketing. It is better integration. Knowing which product, under which conditions, on which field, generates a measurable economic return.

The same dynamic extends into plant nutrition. The math appears straightforward: remove x nutrients, replace x nutrients. But biological systems destabilize that simplicity. A biostimulant increasing nutrient use efficiency changes the economics of the fertility program paired with it. A microbial mobilizing soil phosphorus changes the value proposition of the phosphate application it was intended to complement. The nutrition plan and the biological plan stop functioning as independent decisions. They become coupled systems.

Then comes timing.

A biological product with an optimal application window that does not align with the grower’s operational schedule creates an additional pass: fuel, labor, compaction risk, equipment allocation, and logistical disruption during an already compressed season. For row crop growers, an additional pass during planting or fungicide windows is often measured in hours they do not have. For specialty growers operating around harvest constraints and labor coordination, optimal agronomic timing may directly conflict with operational reality.

The product may work. But if the agronomic timing does not match the operational timing, field logistics win.

And most trial data never measures that calculation.

This is where what I will call crop intelligence enters — not as Silicon Valley branding imported into agriculture, but as operational infrastructure modern agronomy increasingly requires. The industry calls the underlying systems AI, machine learning, or predictive analytics. The grower does not need to care what powers the engine. What matters is whether the system can hold the entire field in view simultaneously.

Consider what a competent biological recommendation now requires. The agronomist must integrate product mode of action, field microbiome conditions, cultivar genetics, weather during the efficacy window, residual chemistry from prior applications, nutrient removal budgets, operational feasibility of another pass, and the economic threshold where the recommendation generates a positive return under that grower’s margin structure.

That is not a recommendation problem. It is a systems integration problem.

And it has exceeded the resolution spreadsheets and static recommendation tables were designed to manage.

Crop intelligence can hold that complexity. Not to replace agronomic judgment, but to make informed judgment possible at modern biological resolution. An agronomist recommending a microbial inoculant from trial summaries and experience on similar soils is making the best decision possible with a partial picture. The same agronomist operating within a crop intelligence workflow — integrating microbiome conditions, nutrient plans, spray schedules, weather forecasts, and operational constraints simultaneously — is making the decision with the full picture available.

Same expertise. Full radar.

The biological and crop intelligence are structurally coupled. The biological creates the complexity. Crop intelligence is the infrastructure capable of managing it at the resolution the biology requires. Neither reaches full potential without the other.

Without crop intelligence, the biological industry will continue producing exceptional science that underperforms commercially — products that succeed in controlled trials, drift toward commodity pricing, and cycle through grower programs because the recommendation infrastructure cannot place them precisely enough to reveal differentiated value.

Without biological complexity to manage, computational systems in agriculture remain a solution searching for a problem sophisticated enough to justify them.

This is the transition now underway.

Not because agriculture suddenly became digital.

Because the products themselves became computationally difficult to manage.

Chemistry simplified decisions. Biology multiplies interactions.

The recommendation infrastructure built around chemistry cannot manage biological complexity at field resolution. Crop intelligence can.

And the grower — the person who prices every recommendation not only in dollars per acre but in labor, fuel, time, logistics, and operational attention — finally receives a recommendation framework capable of matching the complexity of what modern agriculture is asking them to deploy.

The system became misaligned because the products evolved faster than the infrastructure delivering them.

Crop intelligence is the infrastructure catching up.

For the growers and advisors already operating beyond the complexity threshold, the transition has already begun. For the rest of the system, the arithmetic remains the same.

Complexity compounds. Margin compresses. Precision matters.

It is only a matter of time — and margin.

Vatren Jurin, Partner, DunhamTrimmer

The Reckoning publishes monthly Anchors and weekly Field Notes — free, always. Read the full first Anchor, “The Misaligned Machine,” and now the second Anchor “After the Agronomist”https://vatrenjurin.substack.com/p/the-reckoning

Fertility, crop protection, seed genetics, biologicals, micronutrients, adjuvants, timing decisions, and precision recommendations are layered into a single operating system. Each component arrives with evidence. Each was tested, registered, recommended, or sold on the basis of data generated largely inside its own category.

But the grower does not farm categories.

The grower farms the stack.

And the stack — the actual system operating on the acre — is rarely audited as a system.

The first anchors of this publication built the diagnosis. The Misaligned Machine mapped a system designed for a simpler century. After the Agronomist named the trap: a specialist model rewarded for category depth while the biology demands integration across categories. The Field Notes examined the failure points: the grower excluded from design, trial data stripped of context, evidence that outlives its conditions, margin leaking through a chain nobody tracks.

All of that work circled the same object. This piece puts it on the table.

The object is the input stack itself — the full assembly of fertility, crop protection, seed genetics, biologicals, and precision nutrition that a grower layers onto every commercial acre, every season, trusting that the evidence behind each component will hold when the components operate together. This piece audits what is actually underneath that stack. Not product by product. Layer by layer. And then — because this is where the reckoning lives — as a system.

Share THE RECKONING

The Stack Nobody Audits

The specifics vary by crop, region, and operation. But the basic architecture is increasingly common.

A row crop grower building a corn program in 2026 is assembling a nitrogen plan, a fungicide pass, a seed treatment package, a biological inoculant or biostimulant, a micronutrient program, and underneath it all, a hybrid selected for a yield environment and a trait stack that interacts with every other layer in ways no single data set fully characterizes.

A specialty crop grower assembles a different stack with the same structural problem — fifteen to twenty spray applications per season layered onto a rootstock-scion combination or variety selected for pack-out quality, buyer specifications, and market access. The interactions are equally undercharacterized. The per-acre stakes are often higher.

Whether the crop is corn or citrus, the operating reality is similar: an integrated biological system operating on a living substrate under variable weather, managed by a single human being making irreversible decisions under operational pressure.

Every layer arrived with evidence — a trial summary, a university recommendation, a product data sheet, a registration dossier. Each piece of evidence was generated independently, inside its category silo, under controlled conditions, on the genetics commercially relevant at the time. And each was generated largely in isolation from the other layers in the stack.

The grower assembled the layers into a system. Nobody audited the assembly.

A stack audit would not ask only whether a product works in isolation. It would ask whether the assembled program is biologically coherent, economically justified, operationally executable, and still supported by evidence under the grower’s current genetics, soil conditions, pest pressure, labor constraints, market requirements, and risk tolerance.

That is the audit modern agriculture rarely performs.

Layer by Layer

Start with fertility — the oldest, most established evidence base in the stack. Decades of university research, thousands of site-years of data, calibration curves refined across cropping systems. If any layer should be on solid ground, it is this one.

And yet the yield response curves that underpin many recommendations were calibrated on genetics that may no longer exist in commercial production. The soil those recommendations were built on has changed under management — cover cropping altering mineralization rates, no-till stratifying nutrient profiles, regenerative practices transforming the biology the fertility program was calibrated against. In specialty crops, fertigation programs were calibrated on rootstocks that have turned over, under irrigation assumptions that drought years have disrupted.

The 4R framework is sound. The baselines it operates against are drifting. The consequence is not bad science. It is an evidence base that ages quietly beneath institutional credibility, while few incentives exist to revalidate the assumptions underneath recommendations already treated as settled.

Crop protection operates under a different stress. The evidence is not simply aging — it is being actively degraded by the biology it was designed to manage. Resistance evolution has outpaced the evidence cycle in multiple pest systems simultaneously. Mode-of-action rotation strategies were modeled on selection pressure rates that the current intensity of production has exceeded. Fungicide timing recommendations were validated under disease pressure patterns that climate variability has altered.

In specialty crops, the gap carries a sharper economic edge. A grower managing a biocontrol program is doing so under pre-harvest interval constraints, MRL requirements for export markets, residue expectations from buyers, and food safety audit standards the evidence often did not address. The evidence said the product worked. It did not necessarily say the product worked within the regulatory and market framework the grower actually operates under.

Seed is the foundation — the biological platform on which every other input operates. It is also the layer with the fastest turnover and the least integration into the evidence supporting everything stacked on top of it.

A corn hybrid platform turns over every three to five years. The agronomic recommendations layered onto it — nitrogen rates, plant populations, fungicide timing, biological compatibility — were often calibrated on its predecessor. The trait stack compounds this: drought tolerance traits, insect resistance traits, crop architecture, stress physiology, and input response are not independent biological compartments. They interact through plant metabolism and development, yet those interactions remain only partially characterized.

In perennial specialty crops, the problem stretches across decades. A tree fruit grower managing a twenty-year orchard applies a stack whose evidence base was often never updated to match how that genetic platform has aged, how the root system has developed, or how the orchard’s biological baseline has shifted. Whether the cycle is three years or twenty, the failure is similar: the grower builds the stack around genetics using recommendations generated on a different platform, in a different era.

Biologicals — biostimulants, microbial inoculants, biocontrol agents — are the newest addition and the layer most exposed to mismanagement by the existing evidence system. The underlying science is real and increasingly well understood at the mechanistic level. The gap is between the mechanism and the field — between a trial conducted on one soil type, with one genetic background, under one fertility regime, and the commercial acre where every one of those variables is different and interacting.

This is the layer where the failures documented in earlier Field Notes converge: evidence generated under conditions that may not match the field, aging against a shifting baseline, evaluated by a grower who often cannot see the product working because its contribution is physiologically invisible during the decision window.

Biologicals are not primarily a story of failed biology. They are a story of mismanaged integration. A microbial inoculant, biostimulant, or biocontrol agent is often inserted into a program that was not designed around its mode of action, timing requirements, compatibility constraints, or biological dependencies. The product is then judged inside a stack that may suppress, mask, or distort the very mechanism it was built to express.

The biological layer is not being undermined because the biology is inherently wrong. It is being undermined because the evidentiary architecture it inherited was not designed to characterize products whose performance depends on interactions the architecture does not measure.

The Stack as a System

Here is the central failure, and the one no individual layer audit can reveal.

Even if every layer’s evidence were current — recalibrated to today’s genetics, today’s climate, today’s soil biology — the stack would still be unaudited. The evidence for each layer was generated largely in isolation from every other layer. The fertility trial did not account for the biological inoculant. The biological trial did not account for the fertility program. The fungicide timing study did not account for the biostimulant’s effect on plant immune signaling. Each layer was characterized as though it would operate alone. The grower applies them as a system.

The issue is not that every untested interaction is harmful. Many are neutral. Some are beneficial. Some matter only under specific stress conditions. The problem is that the system rarely knows which is which before the grower pays to find out.

Consider a leafy greens operation cycling three crops per year on the same ground. The stack resets every rotation — different variety, adjusted fertility, new pest pressure profile, and biological carryover from the previous crop’s microbial applications still active in the soil. The evidence supporting each input was generated on a single crop cycle. The grower is running stacked cycles where the residual effects of one program become the baseline conditions for the next, and very few trials characterize that compounding in a way that resembles commercial decision-making.

Or consider a high-value vegetable grower whose spray program is fifteen applications deep — fungicides, insecticides, biocontrols, biostimulants, foliar nutritionals — each timed to a growth-stage window measured in days, each interacting with the others in a sequence the evidence base rarely tested as a sequence. The buyer’s quality specifications, the food safety audit, the export MRL — those constraints shape which products can be stacked and when, and most of them were not variables in the trials that generated the evidence.

The row crop grower faces the same void on a longer cycle. Ask any agronomist building an integrated program whether they can point to a data set that characterizes the interaction between the biological they are recommending and the starter fertilizer it will be applied alongside, on the hybrid the grower selected, under the tillage system the grower operates. The answer, with rare exceptions, is no.

The recommendation is built on professional judgment and the assumption that products validated independently will perform acceptably when combined. That assumption has rarely been tested systematically at the full-stack level in commercial agriculture. It is one of the largest unaudited assumptions in the modern input system.

I spent twenty years in product development. Not once — across multiple companies, across fertility, crop protection, biologicals, and nutrition — did I participate in a trial designed to characterize how our product interacted with the full stack the grower would actually apply it within.

We characterized our product. We characterized it well. We did not characterize the system it would enter.

And the way we characterized it revealed the blind spot. The standard trial design tests a new product against the grower standard — the existing program the grower is already running. The question being answered is: does adding our product to the grower’s current system improve outcomes?

That sounds rigorous. It is useful, but incomplete.

The grower-standard trial answers an additive question: what happens when this product is inserted into the existing program? It does not answer an integrative question: what should the program become if this product is truly part of the system?

If a grower adds a microbial inoculant to a corn program, should the starter fertilizer change? Should the in-furrow chemistry change? Should nitrogen timing shift? Should seed treatment compatibility be re-examined? Should placement, moisture management, or early-season stress mitigation be redesigned around the biological mechanism being introduced?

Most commercial trials do not ask those questions. They ask whether the product can survive inside a system built without it. Then they call that a fair test.

The real question — what would the grower’s entire program look like if it were redesigned to integrate this product from the beginning? — is rarely asked. It would require redesigning the trial around the stack. The system is not built to do that.

Every institutional incentive — trial infrastructure, regulatory requirements, commercial positioning — points toward layer-level evidence and away from stack-level evidence.

The grower, once again, inherits the gap.

But the grower does not simply accept it.

The Auditor of Last Resort

There is a practice that every grower I have known engages in and almost none of them talk about publicly.

They test things.

Quietly. Privately. On their own acres, with their own money, on their own terms.

A strip left untreated to see if the biostimulant is doing anything. A few rows where the micronutrient pass is skipped to find out whether the recommendation is returning value. A block where the fungicide timing is shifted a week to see what happens under real conditions. A corner of the orchard where the new biocontrol is applied alongside the existing program and the grower watches — not with a data logger, but with the accumulated pattern recognition of someone who has looked at that canopy every morning for a decade.

These are not published trials. They are not replicated, randomized, or peer-reviewed. They would not survive a statistical methods course. But they are often the closest thing to a stack-level audit that currently exists in commercial agriculture — because the grower is the only person in the system with both the incentive and the operational access to test what the layers actually do when they operate together on their ground.

I did this when I farmed. Every grower I work with now does it. They do it because the evidence the system provides is not sufficient to answer the question they actually need answered: does this stack, on my soil, with my genetics, under my conditions, return more than it costs?

The formal evidence chain cannot fully answer that question. So the grower runs the experiment themselves — undocumented, uncompensated, and invisible to every upstream node that should have provided the answer before the grower had to go looking for it.

The tragedy is not that growers are running these private audits. The tragedy is that the system has made it necessary. And the compounding tragedy is that the knowledge generated — hard-won, field-specific, economically grounded — stays on the operation. It does not flow back to the developer, the researcher, or the advisor. It does not recalibrate the recommendations.

The grower learns something real about how the stack performs on their acre, and that learning dies at the fence line.

The tools to formalize this already exist: structured on-farm trial design, remote sensing, tissue and sap analytics, soil biology assays, metabolomics, microbiome profiling, and AI-assisted synthesis. The barrier is not technological. It is institutional. An audited stack would begin by building the infrastructure to capture what the grower is already learning and connecting it to the analytical tools that can make that learning transferable.

The Record

The input stack the modern grower assembles is the most sophisticated, most biologically complex, and most expensive collection of agricultural technologies ever deployed on a commercial acre. It is also one of the least audited.

Each layer arrived with evidence generated in a different decade, on different genetics, under different conditions, in partial isolation from every other layer. The stack — the system the grower actually operates — is seldom validated as a system. And the only person running the audit is often the grower, alone, at the fence line, with strip trials and observation instead of the tools the system should have built.

That is the record.

Not a record of failure. A record of omission — institutional, structural, and cumulative.

This publication exists to begin correcting it. Not from the greenhouse. Not from the boardroom. From the position of someone who has farmed, built the products, and now audits the system that connects them.

The stack is on the table.

The audit starts here.

Vatren Jurin, Partner, DunhamTrimmer

The Reckoning publishes monthly Anchors and weekly Field Notes — free, always. Read the full first Anchor, “The Misaligned Machine,” and now the second Anchor “After the Agronomist”https://vatrenjurin.substack.com/p/the-reckoning

When I finally got to that development meeting, I understood immediately what was missing from the room: the field.

Not the agronomic field — the data, the trial summaries, the yield maps. Those were on the table. What was missing was the operational field.

The one where a row crop planting window is fourteen days and a late-season wind event does not wait for the crop scout. The one where a vegetable harvest crew is standing in the lane at dawn and a missed fungicide timing does not mean a yield drag — it means a rejected load at the packinghouse. The one where every recommendation arrives as one more decision stacked onto a season that was already overloaded before the first pass was made.

This is not memoir. It is a structural argument.

The grower’s seat teaches specific things about how agriculture actually operates that no other position in the system can. Those lessons look different depending on whether you are managing a corn rotation or a strawberry block. But the fact that the same structural failures appear across both systems is what makes them architectural, not anecdotal.

1. The real cost of a new management step is cognitive, not financial.

In row crops, the pressure is temporal.

The planting window is compressed. Every additional input decision stacks onto a spring that is already saturated: a biological requiring a separate pass, a biostimulant that does not align with the existing spray schedule, or a tank-mix incompatibility discovered three days before planting.

The grower prices these products not only in dollars per acre, but in attention during the most compressed decision period of the year.

In specialty crops, the pressure is sequential and unforgiving.

A vegetable or tree fruit grower may be managing fifteen to twenty spray applications per season, each one timed to a growth stage that does not pause for weather delays or equipment breakdowns. Add a new biological into that rotation and you are not just adding a cost. You are inserting a decision into a sequence where every interval matters.

The penalty for a missed window is not always reduced yield. It may be cosmetic damage that drops the pack-out grade. It may be a residue detection that closes an export market. It may be a harvest delay that collides with labor, buyer commitments, or packinghouse capacity.

The cognitive load is not about one compressed spring. It is about sustained precision across months.

Both systems reveal the same structural failure: the industry prices products in dollars per acre. The grower prices them in attention, logistics, and operational risk.

Those are fundamentally different currencies, and no one upstream is doing the conversion.

This is exactly the kind of field knowledge that would reshape product design if it entered the process at the beginning, instead of arriving later as objection-handling training for the sales force.

2. The product that works and the product that gets adopted are not the same thing.

Efficacy is necessary. It is not sufficient.

In row crops, the barriers to adoption are often logistical. Cold-chain storage the family operation does not have. A formulation that will not tank-mix with the existing program. A value proposition framed as yield advantage that does not survive a conversation with the lender, who wanted to see cost reduction, not incremental gain.

In specialty crops, the barriers are often regulatory and market-driven.

A biological crop protectant with a clean efficacy profile but a pre-harvest interval incompatible with the grower’s harvest schedule is not adoptable, regardless of its trial data. A biostimulant that works beautifully but lacks the documentation a food-safety auditor requires — or that a retail buyer’s sustainability program recognizes — is dead on arrival.

The specialty grower does not just ask whether the product works. They ask whether their buyer will accept it, whether the packinghouse can document it, and whether the export market’s maximum residue limits accommodate it.

Most trials measure none of those things.

The distance between “works” and “gets used” is where a significant share of biological category growth goes to die. In row crops and specialty crops, the failure looks different. But the cause is the same: the trial architecture was built to measure efficacy under controlled conditions, not adoptability under field constraints.

The system has not built the measurement it actually needs.

3. The grower’s most trusted information source is usually another grower.

The data package from the company is information. The neighbor’s experience on the same soil type, under the same weather, in the same rotation — that is evidence.

In row crops, this calibration happens at the coffee shop, at the co-op, and across the fence. In specialty crops, it happens in the packing line, at the grower-shipper meeting, and between operations that share the same pest complex and the same buyer requirements.

The network is different. The function is identical: locally calibrated proof that carries more weight per data point than anything in the formal advisory chain.

This does not mean formal data is irrelevant. It means formal data becomes persuasive only when growers can translate it into local operating conditions.

The industry treats information as flowing top-down. The grower’s actual information network runs horizontally, peer-to-peer, and it operates on a different standard of proof.

Not statistical significance alone. Operational relevance.

Did it work on ground like mine, with constraints like mine, for a buyer like mine?

Any system that treats this horizontal network as noise to be overcome with better data sheets has misunderstood how growers actually make decisions across every cropping system.

4. Every recommendation is a bet the grower makes with their own money.

This is the most structural lesson, and the least discussed.

The product representative makes the pitch and moves to the next account. The agronomist writes the recommendation and moves to the next field. The grower executes the decision and lives with the outcome.

In row crops, that consequence is seasonal: a compressed margin, a difficult conversation with the lender, or a rotation decision that echoes for three years.

In specialty crops, the consequence can be immediate and catastrophic: a rejected load at the dock, a lost retail contract, or a food-safety incident that shuts down a packing operation.

The stakes per acre are orders of magnitude higher. The feedback is faster. The grower’s exposure to a wrong recommendation is correspondingly more severe.

This asymmetry of consequence between the person making the recommendation and the person executing it is the single most underappreciated structural feature of the agricultural input system.

It is not that upstream actors are indifferent. Most agronomists care deeply about their growers’ results.

But the economic architecture does not require them to share the downside.

Care and accountability are not the same thing.

This is why outcome-based advisory is not just a service innovation. It is a response to a structural imbalance.

When margins compress far enough in row crops, and when buyer requirements tighten far enough in specialty crops, growers will demand that the person making the recommendation share the economic consequence of being wrong.

Not as a philosophical principle. As a business model.

The implication is practical.

If the industry takes this seriously, the implications are not abstract.

Product development would begin with operational fit, not just biological activity. Trial design would measure adoptability, not just efficacy. Commercial teams would treat peer validation as infrastructure, not anecdote. Advisory models would move closer to shared accountability, because the grower is already carrying the full consequence of every recommendation.

Seven years in the field did not make me an authority on every crop, every soil type, or every growing region. What it did was show me that the same structural failures — cognitive overload, stripped context, misaligned incentives — appear whether you are managing two thousand acres of corn or forty acres of mixed vegetables.

The specific pressures differ. The architecture that produces them is identical.

Twenty years building products showed me the view from the other side of the same room: smart people designing good inputs inside a system that structurally excludes the field-level constraints that determine whether those inputs get adopted.

Five years auditing the system from the outside confirmed what both seats had already shown me: the gap between the advisory model and the grower’s operational reality is not closing. It is widening, because the products are getting more complex while the delivery infrastructure remains largely unchanged.

The grower’s seat is not the only seat that matters. But it is the seat where every failure in the system arrives as a personal financial consequence — whether that consequence is measured in bushels per acre or boxes per pallet.

That is what makes it irreplaceable as a design input.

Not the grower’s opinion.

The grower’s exposure.

The system was designed from the greenhouse out. The reckoning is that it needs to be redesigned from the field in.

Vatren Jurin, Partner, DunhamTrimmer

The Reckoning publishes monthly Anchors and weekly Field Notes — free, always. Read the full first Anchor, “The Misaligned Machine,” and now the second Anchor “After the Agronomist”https://vatrenjurin.substack.com/p/the-reckoning

Not because it failed. Because they couldn’t see that it worked.

The product didn’t outlive expired evidence. It outlived evidence that was never visible enough to survive the grower’s decision cycle.

The agricultural input system operates on a hierarchy of visibility, and it determines what stays in the program far more reliably than performance data.

When genetics fail, the crop fails visibly. A hybrid that can’t handle disease pressure shows up in stunted stands and firing leaves — visible from the road at sixty miles an hour. When NPK is misapplied, deficiency expresses as chlorosis across the canopy weeks before harvest. When pest control is removed, the damage is physical, obvious, and attributable.

These failures are visible, attributable, and unambiguous. They persist because the cost of removing them is written on the landscape in terms everyone can interpret.

Now consider what a well-formulated biostimulant actually does. It modulates root architecture, upregulates stress response pathways, and improves nutrient use efficiency. The effect is real, but it is distributed across multiple physiological processes — none individually dramatic enough to observe during a mid-season field check.

The same holds for micronutrient and secondary nutrient interventions. And for biological crop protection, the invisibility takes a different form: success appears as the absence of a problem. Pest populations are held below threshold, not eliminated. The crop looks clean. But no dead bugs so the grower cannot distinguish that outcome from a season where pressure never materialized.

These contributions are agronomically and economically real. But during the season—when decisions are made—they are largely invisible.

The proof that this is a visibility problem—not a product problem—is reflected in adoption patterns across cropping systems.

Specialty vegetable systems have led the integration of biologicals, biostimulants, and precision nutritionals for two structural reasons.

First, the feedback loop is compressed. Crops cycle in weeks, not months. A grower can observe treatment effects within the same production window. Evidence does not need to persist—it is continuously regenerated.

Second, per-acre economics support a different risk posture. High revenue per acre allows for iterative testing. A failed trial is absorbable.

Row crops and perennial systems sit at the opposite extreme. Feedback loops extend across a full season—or multiple years. Margins compress tolerance for inputs that cannot be verified in-season. The same product that earns a stable position in a vegetable program cycles in and out of a row crop program because neither condition holds.

The products are not different. The evidence architecture is.

The industry often describes this as a retention problem. It is not.

It is an evidence problem wearing retention’s clothes.

The advisory system compounds it. Agronomists operating under volume-based compensation models do not have the time, incentive, or economic structure to build and communicate a product-specific evidence case for something the grower cannot directly observe. Evidence generation is not monetized. So it does not happen.

Recommendations persist by inheritance rather than validation—until grower skepticism removes them.

The limiting factor for biological category growth is not efficacy. It is evidence visibility.

The tools to address this gap already exist.

In-season tissue metabolomics can detect physiological shifts weeks before visual expression. Root imaging can quantify below-ground effects that scouting cannot access. High-resolution remote sensing, when calibrated to mode of action, can detect treatment response at field scale.

The constraint is not tool availability. It is integration.

These tools are not embedded in the decision pathway that connects recommendation to evaluation. The agronomist cannot show the grower what the product did before the grower decides whether to keep it.

The product that outlives its evidence is not a failed product. It is a working product inside an evidence system designed to detect failures, not to visualize contributions.

Until that system evolves, adoption will continue to be governed by visibility—not performance.

That is not a science problem. It is a systems problem.

And it is the kind of problem this publication was built to name.

— Vatren Jurin, Partner, DunhamTrimmer

The Reckoning publishes monthly Anchors and weekly Field Notes — free, always. Read the full first Anchor, “The Misaligned Machine,” and now the second Anchor “After the Agronomist”https://vatrenjurin.substack.com/p/the-reckoning

In the first issue of The Reckoning, I laid out a map. The agronomic system delivering modern biological inputs was designed for a different century—siloed by category, optimized for throughput, and disconnected from the grower’s economic reality.

Since then, we examined that misalignment across three points: the product development chain, where the grower enters last; the evidence layer, where data travels without context; and the knowledge architecture itself, where complexity exceeds individual capacity.

That was the diagnosis. This is what comes next.

The word agronomist is used as though it describes a single role. It does not. It describes at least three distinct functions operating at different points in the system.

The advisor—the crop consultant, the CCA, the dealer agronomist—translates product knowledge into field-level recommendations and stands across the table from a grower making a decision that cannot be undone after planting. The developer—the product manager, formulation scientist, or field development lead—builds the input, designs the trial program, and determines what the advisor has to work with. The research scientist—the discovery-stage investigator, extension specialist, or R&D lead—generates the next generation of tools and defines what the developer will build years from now.

These are fundamentally different jobs, requiring different training, different decisions, and different measures of success. Yet the industry collapses them into a single label and constrains them within the same system. I have worked across each of these roles. The constraint is consistent.

The Generalist Problem

There is a phrase that has followed generalists through every industry for as long as specialization has dominated: jack of all trades, master of none. It is meant as a dismissal. In practice, it describes the most undervalued structural role in modern agriculture.

Specialists see their domains with extraordinary clarity. Fertility agronomists understand nitrogen dynamics with precision earned over years. Crop protection advisors manage resistance, timing, and mode-of-action rotation with depth that cannot be improvised. Seed specialists translate trait stacks and performance data into recommendations calibrated to specific yield environments. These are serious professionals doing serious work, and nothing in what follows diminishes that.

But the specialist, by definition, sees their category—not the gaps between categories. And in a system where the biology is now integrated, those gaps are where value is lost. A V4 biostimulant can alter plant signaling and change a fungicide decision at R1. Inoculant performance depends on a fertility program it was never designed alongside. New genetics express differently under nitrogen regimes calibrated for previous generations.

The generalist does not replace the specialist. The generalist sees what the specialist cannot: how decisions interact once they converge on the same plant, in the same field, under the same conditions. This is not shallow knowledge. It is a different kind of mastery—the mastery of interactions rather than categories, of systems rather than components. It is also precisely the capability the next generation of agriculture demands, and the current model is not built to produce.

A System Designed for Specialists

The agricultural system—from university training to corporate structure to retail incentives—was built to produce specialists and reward them for staying specialized.

Agronomy education is organized by discipline. Soil science, plant pathology, entomology, weed science, and crop physiology are taught as distinct domains. Each is deep, rigorous, and necessary. But the student who crosses boundaries gains no structural advantage over the one who remains within a single discipline.

Corporate structures reinforce the same logic. Product managers own product lines. Fertility teams do not sit in biological pipeline reviews. Crop protection leaders do not operate with visibility into seed trait timelines. Scientists publish within disciplines and are evaluated on contributions to their pipeline, not on interactions across pipelines that no one assigned them to study.

The incentive architecture is vertical: go deeper, publish more, capture more share within your category. The horizontal view—how inputs interact in the field—is not just unrewarded. It is structurally invisible.

The advisory layer replicates this pattern. The consultant attempting to build an integrated program—across fertility, crop protection, and biologicals—faces fragmented data, category-based compensation, and time that no one is paying for. The system does not require bad actors to produce poor integration. It requires only incentives that reward depth within silos over synthesis across them.

That model worked when interactions were limited, conditions were stable, and margins absorbed inefficiency. That era is over.

What Changed

The shift was not singular. It was convergent.

First, biological complexity increased. A modern biological program is not a product; it is a system of interacting agents whose performance depends on multiple variables simultaneously:

• Soil microbiome composition

• Plant genetic background

• Environmental conditions at application

• Residual chemistry from previous seasons

• Timing and sequencing of other inputs

As a result, fertility, biological, and crop protection decisions are no longer independent. They are coupled—biologically, economically, and temporally.

At the same time, the data layer expanded. Precision agriculture platforms now generate sub-field prescriptions from satellites, drones, and sensor networks. Equipment produces real-time as-applied data. In-field sensors measure soil and canopy conditions continuously rather than periodically.

At the R&D level, the acceleration is even more pronounced. Metabolomics reveals interaction dynamics at a biochemical level. CRISPR compresses trait development timelines. Bioinformatics processes genomic and metagenomic data at scales that allow functional mapping of soil and plant responses.

Each of these advances is valuable independently. Together, they create a compounding data problem.

Spatial, temporal, efficacy, interaction, and genomic data all exist—but they live in separate systems, interpreted by different specialists, and rarely synthesized into a single decision framework. The information is not missing. The integration is missing.

This is where the generalist becomes essential—not as a philosophical preference, but as a structural requirement. No specialist can hold all of these dimensions in a single frame. But the system now requires exactly that.

The Transition

The full proverb says it clearly: jack of all trades, master of none—but oftentimes better than a master of one. Agriculture dropped the second half. At the exact moment when biology and data demand integration, the system continues producing specialists who are not trained, incentivized, or positioned to operate across boundaries.

The question is not whether the transition will occur. It already has. The question is how—and who leads it.

There are two paths.

The first is the human generalist: individuals who have deliberately built cross-domain experience and understand the system rather than just the category. They exist, but not at scale, and rarely with institutional support.

The second is the technological generalist: artificial intelligence operating as a synthesis layer. An AI-assisted workflow does not replace the specialist. It places their expertise inside a larger context—connecting nitrogen decisions to biological performance, aligning application data with microbiome responses, linking fungicide timing to plant signaling, and integrating all of it with the grower’s economic constraints.

The agronomist still makes the decision. But for the first time, that decision can be made with the full picture rather than a partial view.

This is not theoretical. It is operational. Tools that were non-functional eighteen months ago are now performing analytical work that previously required coordinated teams—if that coordination happened at all. One generalist, augmented by AI, can now operate with a level of integrative capacity that the current system does not produce institutionally.

That is not a future state. That is today.

Who Moves First

The transition will not be led by large input companies. Their structures, incentives, and product-line economics are built on specialization. Integration challenges those foundations. They will adapt eventually, but they will not lead.

The university system will not lead either. Academic incentives reward disciplinary depth, not integration. Cross-disciplinary work remains harder to fund, publish, and build a career around.

The transition will instead be led by those already operating at the intersections: advisors building outcome-based models, developers redesigning product systems around real-world programs, researchers integrating field data with biological insight, and ultimately growers.

Growers are the true integrators. Every decision they make is a systems decision, whether the upstream chain acknowledges it or not. As margins tighten—and they are tightening—the grower will shift from paying for category-level advice to paying for integrated recommendations that return dollars per acre.

That economic pressure will force the transition. Those who recognize it early will not win because they have better products or more data. They will win because they operate with a better model.

After the Agronomist

After the agronomist is not a world without agronomists. It is a world where the advisor, the developer, and the scientist operate inside a system that matches the complexity of the agriculture they serve—where biological data, precision layers, genomic insight, sensor networks, and economic constraints are synthesized rather than siloed.

The specialist is not obsolete. The specialist operating in isolation is.

The system trained us to go deep. The biology requires us to go wide. The data demands that we do both.

The people who can—or who build the tools that make it possible—will define what the next generation of agriculture delivers.

This is the transition.

It is not coming. It is already here.

— Vatren Jurin, Partner, DunhamTrimmer

The Reckoning publishes monthly Anchors and weekly Field Notes — free, always. Read the full first Anchor, “The Misaligned Machine,” and now the second Anchor “After the Agronomist”https://vatrenjurin.substack.com/p/the-reckoning

Share THE RECKONING

Human beings have been building external memory systems for as long as we have faced decisions too complex and too consequential to hold in a single mind. Cognitive scientists have a word for this: exogram. An exogram is simply knowledge pushed outside the skull and into the world — marks on bone, notches on ivory, signs carved into stone. Not decoration. Not ritual. A tool. The external storage of structured knowledge that allows decisions to survive the distance between one season and the next, between one person's expertise and the operation that depends on it.

A study published earlier this year in *PNAS* — Bentz & Dutkiewicz, "Humans 40,000 years ago developed a system of conventional signs" — analyzed geometric signs carved onto portable objects from roughly 40,000 years ago, some of the oldest deliberate markings we have found.¹ The finding was not that these signs were writing. It was something more foundational: the signs show measurable, repeating structure consistent with a deliberate shared code. Repetitive. Compressed. High signal-to-noise. The people who made them were not recording stories or language. They were likely recording structured, reusable information — the kind of knowledge that needs to survive weather, seasons, and the death of the person who originally held it in their head.

That is an agronomist's problem. It has always been an agronomist's problem.

When early agricultural societies emerged, the cognitive challenge did not change. It scaled. Managing a seasonal food system generates a decision load that exceeds what any individual memory can reliably hold. The first writing systems in Mesopotamia did not emerge to record poetry or history. They emerged to manage grain inventories and track inputs against outputs. The clay tablet was not a literary invention. It was a farm management tool — the exogram that stabilized the first large-scale agricultural economies.

Every advance in agricultural knowledge management since has followed the same logic. The field notebook. The soil test. The yield monitor. The agronomic database. Each one a better exogram — a new way of storing what matters outside the individual so that decisions can be made with the full picture, not just the fraction any one person can carry.

Here is what makes the current moment different: for the first time, the exogram can talk back.

Large language models — the technology at the core of tools like ChatGPT, Claude, and the AI systems now entering agricultural workflows — are trained on vast bodies of text, research, trial data, and accumulated human knowledge. What they do, at the most basic level, is predict: given everything that has been encoded before, what is the most useful next step? That is not magic. It is pattern recognition operating at a scale no individual agronomist can match — synthesizing across soil science, biological research, weather data, and agronomic experience simultaneously, in the time it takes to ask a question.

The biological complexity of today's input programs — live microbial products interacting with soil ecosystems, new genetics expressing differently under stress, multiple inputs whose interactions have never been fully characterized — has outgrown the exogram we currently have. The field notebook, the agronomist's memory, even the conventional database: none of them were built to hold this level of complexity in a form that is actually usable at decision time.

What AI makes possible — for the first time — is an exogram that does not just store agronomic knowledge. It actively synthesizes it. An advisor running AI-assisted workflows can cross-reference a biologicals trial dataset against regional soil health indices, scan a decade of interaction literature, and generate a draft recommendation in the time it previously took to find the right spreadsheet. That is not replacement. That is leverage — and it is the only rational response to a decision load that has outgrown its tools.

The farms, advisors, and companies that understand this first will not be replacing agronomists. They will be doing what every agricultural civilization has done when the decision load outgrew the available memory system: **they will build a better exogram.

The algorithm is not arriving from outside agriculture. It is the latest version of something agriculture invented forty thousand years ago: a better way to remember what it cannot afford to forget.

--- ¹ Bentz, C. & Dutkiewicz, E. (2026). Humans 40,000 y ago developed a system of conventional signs. *Proceedings of the National Academy of Sciences.* https://doi.org/10.1073/pnas.2520385123

That reckoning is coming.

Subscribe now

— Vatren Jurin, Partner, DunhamTrimmer

The Reckoning publishes monthly Anchors and weekly Field Notes — free, always. Read the full first Anchor, “The Misaligned Machine,” at https://vatrenjurin.substack.com/

I’ve run a lot of trials. I’ve also read a lot of trial reports written by people who ran trials on their own products.

The conditions aren’t a footnote. They’re the product.

Here is something those reports almost never say plainly: the trial was designed to answer a specific question, under specific conditions, at a specific moment in the crop’s development — and the answer it produced is only valid inside those parameters.

What gets published — and more importantly, what gets put in the sales deck — is the number. The percent response. The yield advantage over the untreated check. Stripped of its conditions. Presented as though it will transfer to every field, every soil type, every application timing, every agronomic program it encounters.

It won’t. And everyone who designed the trial knows that.

The site was selected because the soil type responds to the product’s mechanism. The application timing was optimized across years of preliminary work before the formal trial ran. The untreated check was managed to keep it alive — not to reflect what a grower’s actual baseline program produces. The unfavorable year got attributed to “anomalous weather” in the footnotes.

None of this is fraud. It is how product development works. The problem is the next step.

The conditions are not a footnote. They are the product.

When data leaves the development team and enters the commercial chain, the conditions get dropped. But those conditions — soil organic matter threshold, application window relative to growth stage, temperature range during efficacy expression, crop water status, baseline fertility level — are not background noise. They are the operating manual. They tell you exactly when and where this product will perform, and by implication, exactly when and where it won’t.

A grower who understands the trial conditions can make a decision. They can ask: do my conditions match what the trial was measuring? If yes, the data is meaningful guidance. If no, the data is someone else’s result.

That question is almost never asked — because the conditions were never communicated. The number travels. The context doesn’t.

Communicating conditions doesn’t undermine a product. It builds the trust that creates durable commercial relationships. A grower who knows when your product works, and when it doesn’t, will use it correctly, see the result you characterized, and come back. A grower who applies it under the wrong conditions, sees a fraction of the published response, and blames their own execution is a grower who quietly stops buying.

The statistical framework we built our evidence in

There is a deeper problem. The statistical models used to design and analyze field trials were built for a world with few variables — randomized complete block design, analysis of variance, treatment means. Frameworks developed when inputs were simpler and interacting variables were manageable enough to control for statistically.

That world no longer exists on a commercial farm.

A modern biological input interacts with soil microbiome diversity, plant genetic background, weather patterns too localized for standard station capture, residual chemistry from last season, and a dozen other variables that are either not measured, not measurable with standard trial infrastructure, or excluded from the model because including them would require a trial size no company can fund.

So we run the old model. We report the old metrics. We present results as though the variables we didn’t measure don’t exist.

The consequence is a systematic underestimation of why products behave differently across environments — and a systematic overestimation of how transferable any single trial result actually is. The variables we couldn’t measure in 1985 are now measurable. Soil health metabolomics, plant stress biomarkers, microbiome profiling, hyperspectral imaging — the tools exist. They are simply not integrated into the trial design frameworks the industry uses to build its evidence base.

We are reporting a new generation of products through the lens of a statistical architecture built for the previous one. The number survives. Everything that would make it useful — and honest — does not.

The blind spot isn’t ignorance. It’s a choice to keep measuring what we’ve always measured.

The tools to do this differently exist. The willingness to admit the old model is no longer adequate does not — not yet.

That reckoning is coming.

— Vatren Jurin, Partner, DunhamTrimmer

The Reckoning publishes monthly Anchors and weekly Field Notes — free, always. Read the full first Anchor, “The Misaligned Machine,” at https://vatrenjurin.substack.com/p/the-reckoning

I’ve sat in a lot of rooms the grower never sees.

The product launch meeting where we set the positioning and the trial data we’d lead with — and the trial data we wouldn’t. The internal review where someone asked whether the new biological would perform in drought stress conditions and the answer was “we don’t have that data yet” and the launch timeline didn’t move. The sales training where we taught a room full of agronomists how to handle objections before they’d ever handled the product in a real field.

Those rooms are where the agronomic system gets built. The grower is never in them.

The problem isn’t intent. The problem is sequence.

The grower enters the development process in the last third — almost always at the field trial stage, when the product already exists, the formulation has been decided, the label is drafted, the packaging is fixed, the price is set, and the value proposition has been written by people whose primary data source was the research greenhouse and the competitive landscape spreadsheet. The grower is asked one question at that stage: does it perform?

Performance is real. It matters. But it is a fraction of what a grower actually brings to a product’s commercial success or failure.

Take formulation. Whether a biological is delivered as a liquid, a wettable powder, a seed treatment, or a granule isn’t just a manufacturing decision — it determines whether the product integrates with the grower’s existing equipment and tank mix program, or gets left on the shelf after the first application. Packaging determines whether it fits in a 40-foot equipment shed with specific shelf constraints, or requires cold chain infrastructure that a family operation simply doesn’t have. The value proposition is what a grower has to carry into a conversation with their lender when they add a new line item to the input budget. And differentiation — the hardest thing to build in a market full of products making similar claims with similar data packages — never gets built from a greenhouse.

None of these questions get asked when the grower is only present for the performance chapter.

And the grower has answers. Substantive, specific, commercially precise answers. Because the grower is the steward of the crop from the moment the seed goes in the ground until the combine comes out of the field. They understand the constraints — agronomic, logistical, financial, relational — better than anyone upstream in the development chain. The grower knows which formulation they’ll actually use versus the one they’ll leave on the shelf after the first application. They know which claims will get a second look from the agronomist at the co-op and which ones get filed in the mental category of “we’ll see.” They know what the real cost of a new management step is — not in dollars per acre, but in cognitive load during a planting window when sixteen other decisions are already queued.

That knowledge doesn’t enter the development process. It enters the objection-handling training.

By the time a recommendation reaches the field, it has passed through a chain that looks like expertise but functions like a filter. The DSM filters for what the company needs to move. The dealer filters for what keeps the relationship intact. The agronomist filters for what they can defend at the next crop walk. Each layer adds a commercial consideration. Each layer removes a little of the uncertainty that was sitting in the original trial data. The grower gets the output of all that filtering — and inherits whatever was lost along the way.

This is what I mean when I say the machine is misaligned. It is not that any single person in that chain made a bad decision. It is that a system with the grower at its nominal center has architecturally placed the grower at its operational end. The person with the most at stake — land, capital, crop, livelihood — is the last stakeholder consulted in building the programs they will ultimately execute.

I know what those upstream rooms sound like. I was in them for twenty years. I came to them from the field — which is how I learned what they were missing.

I also know what it feels like to be the one holding the sprayer at the end of a recommendation chain I didn’t design, making a call I can’t undo.

The misalignment isn’t abstract. It lives in that gap — between the room where the program was built and the field where it has to perform.

Move the grower to the beginning. Not as a focus group. As a design partner. That means a grower in the formulation review meeting, not the launch training. A trial program designed around the constraints they actually face, not the conditions where the product performs best. A value proposition built from a lender conversation, not a greenhouse. That is a different kind of commercial agriculture.

That is also the reckoning that is coming — whether the industry chooses it or has it chosen for them.

Subscribe now

— Vatren Jurin, Partner, DunhamTrimmer

The Reckoning publishes monthly Anchors and weekly Field Notes — free, always. Read the full first Anchor, “The Misaligned Machine,” at

THE RECKONING

My personal Substack

By Vatren Jurin

.

Less than 1% of the world’s population operates the commercial farms that sustain the modern global food supply.

It is an extraordinary logistical achievement in human history. And today, we are asking that tiny fraction of humanity to do more—with less margin, less water, less labor, and less time than ever before.

Let’s start with the numbers that should reframe every conversation in this industry.

Somewhere between 570 and 600 million farms operate across the planet. The vast majority of those are smallholder or subsistence operations. The commercial operations that actually feed the global urban population at scale—the farms producing the surplus calories, protein, and fiber that power the global supply chain—represent a fraction of a fraction of humanity. Less than 1% of the world’s population has assumed the commercial production burden for the rest of us. In the United States alone, just 1.9 million farms produce food for 335 million Americans, plus a substantial portion of the global export market.

That is not just a statistic. That is a civilization-scale dependency built on an extraordinarily thin layer of human expertise, biological knowledge, and agronomic judgment.

And we are asking that layer to do something it has never been asked to do before.

The generation of crops being developed and deployed right now—stress-tolerant varieties, nutrient-efficient genetics, biologicals engineered to interact with the plant biome in ways we are only beginning to understand, precision inputs calibrated to the square meter rather than the field—represents a fundamentally different kind of agricultural system than the one that got us here.

The agronomic complexity is greater. The data requirements are higher. Environmental and regulatory pressures are higher. The decisions are far more interdependent. The margin for error, in an era of volatile input costs and compressing farm economics, is thinner than at any point in the last fifty years.

The potential is real. So is the gap between what this new generation of agriculture is capable of and what the system currently delivers to the farmer who has to make it work by planting time.

That gap is what this publication exists to close. And to understand it, you have to understand something most people in the industry would prefer not to say plainly:

The agronomic system that will need to serve this new generation of crops was designed for a different century. It is not broken in any one place. It is misaligned across every place simultaneously.

I’ve watched this from three seats:

• Seven years as a grower—actually in the field, managing inputs, making decisions on the line.

• Twenty years as a product manager—building and launching inputs across the full agronomic stack inside companies large enough that a single product decision affected millions of acres.

• And now five years as a consultant—sitting across from the executives, investors, and agronomists trying to figure out what comes next.

What I’ve seen from all three positions is the same thing, expressed differently depending on where you sit: the people in this industry are largely doing their jobs well. The system those people operate inside is not doing its job.

At the exact moment when agriculture needs to level up—when the new generation of crops demands a more sophisticated, more integrated, more economically precise system to reach its potential—the misalignment is becoming more expensive, not less.

This is The Reckoning. And this first piece is the map.

The New Crop Doesn’t Care About Your Old Program

The biological complexity of the current generation of crop inputs is genuinely unprecedented.

A modern biologicals program involves live microbial organisms competing for rhizosphere real estate with native soil populations. A new nutrient-use-efficient variety expresses differently under stress than its predecessor—a complication magnified by changing weather patterns that are introducing entirely new abiotic and biotic stress behaviors into the field. A biostimulant designed to upregulate a plant’s defense signaling may interact with a fungicide applied three weeks later in ways that neither the biostimulant company nor the fungicide company has characterized. A precision variable-rate fertility prescription generated from satellite imagery and yield monitor data is built on assumptions about crop response curves that were calibrated on genetics that no longer exist in the field.

None of this is theoretical. It is happening on commercial farms right now, at scale, without the agronomic infrastructure to manage it.

The problem is not that the new inputs are bad. Many of them are genuinely remarkable, representing decades of biological research and significant capital investment. The problem is that the system delivering them to the field was built to handle a simpler kind of agriculture.

Fertility, crop protection, seed—distinct categories, distinct advisors, distinct decision processes, largely independent of each other. That model worked reasonably well when the interactions between inputs were relatively few, the climate was more predictable, and the margin for error was wider.

It does not work when the input program is a complex biological system whose components interact at the molecular level, and whose performance is highly sensitive to environment, timing, and sequence.

The fertility advisor recommends nitrogen, phosphorus, and potassium based on a soil test and a yield goal established before the new variety was even released. The biological rep positions the inoculant as a complement to the fertility program without access to the fertility data. The crop protection program is built around resistance management protocols with no reference to what the biological inputs are doing to the plant’s immune state. The precision ag platform captures the yield result at the end, but cannot attribute it to any specific element of the program that produced it.

Nobody in that system is wrong. The integration of all of it has simply never been attempted. And for the new generation of crops—which are more responsive to environment and management than any generation before them—the cost of that non-integration is measurable in dollars per acre, every season.

The Profitability Problem Nobody Is Naming

Here is the economic reality that sits underneath all of this, and that the industry is remarkably reluctant to discuss with the directness it deserves.

The commercial farmers who feed the world do not do so charitably. Farming is a business. The operations that produce the calories the global population depends on operate on margins that would make executives in almost any other industry deeply uncomfortable.

Input costs consume a substantial portion of every revenue dollar. Weather, markets, and regulatory environments introduce variability that cannot be fully hedged. The financing of land, equipment, and operating capital requires a level of economic precision that most input programs are simply not designed to support.

Now, layer on the external squeeze. Today’s commercial growers are operating under regulatory-guided policies dictated by consumers rather than producers. The immense, top-down pressure for “sustainability”—often defined in corporate boardrooms and enforced through supply chain mandates—is being placed squarely onto inherited agronomic systems that were built purely to maximize yield, not to execute complex ecological accounting. The farmer is being asked to shoulder the financial and operational risk of the world’s sustainability goals using an advisory infrastructure that wasn’t designed for the job. You cannot overlay a mandate for regenerative, biologically complex farming onto a siloed, legacy framework and expect the math to work.

And yet the agricultural input industry—which exists to serve those farmers—is largely organized around product performance claims and category-level marketing rather than farm-level economic outcomes.

A biostimulant is sold on percent-yield response data from replicated trials at university research stations. A new seed trait is priced against its theoretical yield advantage in high-potential environments. A precision ag platform is marketed on the number of data layers it ingests rather than the number of profitable decisions it has actually enabled.

The farmer sitting across the table from a product rep, an agronomist, and a data platform is trying to answer one question:

“Will this input pay for itself on my operation, in my soils, in my rotation, against this year’s weather and compliance burdens, at current commodity prices?”

That question—specific, economic, local—is almost never what the product is designed to answer.

The new generation of crops changes this calculation in two directions simultaneously. On one hand, the biological upside of these varieties and inputs is real; the potential to produce more food with fewer synthetic inputs, on more marginal soils, under more variable weather conditions, is not marketing. It is grounded in legitimate science.

On the other hand, the complexity and cost of realizing that potential is higher than the industry is currently equipped to support. The gap between what a new biological system can do in a well-managed program and what it delivers in an average-managed program is wider than it has ever been.

That gap is profit leaking out of the farm and back into the input supply chain—not because the inputs are bad, but because the system connecting input to outcome was not built for this level of complexity.

The Intelligence Layer We Actually Need

There is a third force now entering the agronomic system, and it is moving faster than most of the industry is ready to absorb.

Artificial intelligence—specifically, large language models and machine learning systems trained on agronomic, genomic, and environmental data—is beginning to do something the agricultural inputs industry has never had access to before: it is making the complexity of the new biological system manageable at the farm level.

I am building AI-assisted workflows into my own consulting practice right now. Not as an experiment. As the primary operating model for how I analyze trial data, synthesize market intelligence, translate biological research into agronomic recommendations, and communicate complex system dynamics to clients who need to make decisions by next week. The acceleration is not theoretical. Tools that were not functional eighteen months ago are now doing substantive analytical work that would previously have required a team.

For the agricultural system, the implications are significant and underappreciated. The reason the new generation of crops has not yet fully delivered on its biological potential is not a science problem. The science has largely been solved. It is an information integration problem. The agronomic system does not currently have the capacity to manage the complexity of a biological input program at the farm level, in real time, across the shifting variability of soils, climates, and regulatory constraints that commercial agriculture encompasses.

AI changes that. Not by replacing the agronomist or the consultant or the grower’s judgment—but by giving those people access to the synthesis, the pattern recognition, and the decision support that the compounding complexity of the new agricultural system actually requires.

The farms, companies, and advisors who understand this first will operate with a structural advantage over those who do not. The alignment problem described in this piece—between the sophistication of the new inputs, the shifting environment, the regulatory pressures, and the capacity of the system to deploy them all—has a solution. It is beginning to be built. And the people in the room when it gets built will determine what the next generation of agriculture actually delivers.

What The Reckoning Is For

Less than 1% of humanity carries the commercial production burden for the rest of the modern world. That fact deserves more than a passing acknowledgment. It deserves a level of seriousness about getting the agronomic system right—about aligning the intelligence, the economics, and the biological capacity of modern agriculture with the scale of what it is being asked to do.

This publication is written from a rare vantage point where that full picture is visible: someone who has grown the crops, built the inputs, and now sits at the table where strategy gets made. What follows, month by month, will be an honest accounting of where the system works, where it fails, and what the path forward actually looks like for the farms, companies, and investors who are serious about the answer.

No advertisers. No products to protect. No distribution channel to keep happy.

• If you’re a grower navigating a new generation of inputs with an old generation of advisory infrastructure—you’re in the right place.

• If you’re building or investing in the companies trying to solve the alignment problem—you’re in the right place.

• If you’re an agronomist or consultant trying to understand how the next decade reshapes your role—you’re in the right place.

The machine is misaligned. The reckoning is already underway. And for the first time, we have the tools to fix it.

Let’s get to work.

Vatren Jurin

Partner, DunhamTrimmer

Subscribe now

The Reckoning is published monthly, with weekly Field Notes between issues. It is free, always. If this piece made you think differently about something, forward it to one person who needs to read it.

Vatren Jurin, Partner, DunhamTrimmer