The dominant pattern in agtech for the last fifteen years has been best-of-breed. Pick the best spray diary. Pick the best livestock tracker. Pick the best accounting package. Pick the best compliance reporter. Pick the best workforce tool. Connect them with CSV exports, webhooks, and the occasional Zapier flow. The pitch is that you get the best of every category, while paying the per-app subscription that gets you what you actually need and nothing more.
It's a pitch that makes intuitive sense. It mirrors how most other industries assemble their software stacks. And in industries where the data flows are mostly one-directional and the records mostly stand alone, it works.
It doesn't work in agriculture. It has never quite worked in agriculture. The structural reasons it doesn't work are not obvious from inside any one of the individual best-of-breed products, all of which are usually excellent at what they do. They become obvious only at the seams — where the spray diary should be the same record as the audit pack, the same record as the financial entry, the same record as the buyer's chain-of-custody log, the same record as the worker's certification check. The best spray diary on the market is still not those other things, and nothing the spray diary's vendor can do will make it those other things.
This piece is the case for that observation. Not against any particular best-of-breed product. Against the model.
Why best-of-breed works in other industries
It's worth being precise about where best-of-breed does work, because the parallel is instructive.
In most modern software stacks — particularly in office, sales, marketing and engineering teams — best-of-breed works for two reasons. First, the records are usually independent. The CRM doesn't need to be the same record as the helpdesk ticket; they reference the same customer but they describe different events. Second, the integrations that bridge them are mature, well-funded, and largely unidirectional — a webhook fires when a deal closes, a downstream tool updates a status, and nothing further needs to happen.
When records are tightly coupled — engineering, for example — the industry has trended toward platforms (GitHub, GitLab) that bundle source control, code review, CI, issue tracking, deployment and release management into a single ledger. Even there, the best-of-breed model lost.
Agriculture is closer to the second pattern than the first. The records aren't independent. A spray applied to Paddock 14 is a compliance event, is a financial event, is a withholding-period event, is a worker-certification event, is a buyer chain-of-custody event, is a lender treatment-register event. The same record. Different consumers. Best-of-breed treats these as separate records that need to be synchronised. Reality treats them as one record that needs to be in many places at once.
Three things that go wrong reliably
Best-of-breed agriculture stacks fail in three predictable ways. None is a defect of any individual product; all three are consequences of the model.
Records drift apart. When the spray diary lives in App A and the financial entries live in App B, the version that ends up in the audit pack is whichever was re-keyed most recently. Every gluing point — every CSV download, every webhook, every “let me just paste this in” — is a place for the two records to disagree. By the time the auditor or the buyer asks, you're not sure which version is true.
The agtech vendor's response to this is integration. The customer's experience of integration is: it works until it doesn't, the breakage is silent, and the discovery of the breakage usually happens during the audit. The total cost of “best of breed, stitched together” is not the per-app license bill. It's the licence bill plus the hours spent re-aligning the data, plus the deals not closed because the proof wasn't there, plus the compliance moments that cost weeks of preparation instead of one click.
The seams cost you the value chain. Your bank can't see your livestock register. Your buyer can't see your compliance status before they bid. Your processor can't push the settlement entry into your accounting. Your agronomist can't see whether their prescription was applied as recommended. None of the stitched-together apps were designed to talk to your lender, your buyer or your processor — so they don't. The friction between your operation and the people you transact with is the friction between your apps.
This is the part of the best-of-breed argument that's quietly fatal. The point of running good operational software is not so the farmer can see their own data more easily; it's so the entire value chain can. Best-of-breed structurally fails this because no vendor in the chain has the incentive to make their tool the system of record for someone else's relationship.
Every integration is a relationship you maintain. When App A pushes an update and breaks the CSV format, you find out on the morning you're trying to generate an audit pack. When App B's API quota gets exceeded, you find out when the data stops flowing — usually a few weeks after the flow stopped, when someone asks why a report is empty. When App C's vendor gets acquired and rewrites the schema, you find out when the integration silently stops. Multiply by the seven apps in the typical farm software stack and you're maintaining 21 bilateral relationships, none of which are your day job.
Why the agtech market produced this in the first place
Best-of-breed isn't an accident of the agtech market. It's the natural outcome of how SaaS economics work.
You can fund a small team to build the world's best spray diary, charge $50 a month for it, and have a viable business. You can fund a small team to build the world's best livestock tracker, charge $80 a month, and have another viable business. You can keep doing this for each category — workforce, compliance, accounting, marketplace, IoT, driver app — and end up with seven viable businesses, none of which are the operating system for agriculture, all of which are best-in-class within their slice.
You cannot fund a team to build the world's best operating system unless you're willing to underwrite the much-longer arc it takes to get the connected piece right. The build is longer, the customer payback is slower, and the market keeps telling you that the best-of-breed model is fine because, look, every individual category has a viable product.
The economics produce the structure, the structure produces the customer experience, and the customer experience is: my records exist; they're just not in any single place.
When best-of-breed actually works in agriculture
It's not zero. Three cases where the best-of-breed model genuinely serves the operation:
- A single dominant enterprise with simple compliance. A small cropping-only operation with no livestock, low compliance overhead and no value-chain integration requirements can run on a single excellent crop tool plus accounting and be fine. The seams don't matter because there's only one significant seam.
- Highly specialised operations where one tool is irreplaceably better. Some genuinely deep precision-ag tools — yield monitoring, variable-rate prescription, satellite-derived imagery — are so much better than the generic versions in unified platforms that the cost of the seam is worth it. The right answer there is usually to use the deep tool plus a unified platform that ingests its outputs.
- Operations where the operator is also the systems integrator. If you happen to be an operator who genuinely enjoys maintaining seven integrations and can spend half a day a week on it, you'll get more bang per dollar from best-of-breed than from a unified platform. This is a small subset of operators.
In every other case, the best-of-breed model is more expensive than it looks. The expense is just hidden in your time, your missed contracts and your audit failures rather than your software bill.
What the alternative actually looks like
The opposite of best-of-breed is not “one terrible app that does everything badly.” The opposite of best-of-breed is one operating system that ships primitives — the verbs every farm operation uses — and lets the features fall out of them.
An operating system doesn't ship a spray diary. It ships the verb: record a chemical application against this paddock, with this delta-T and this wind reading, with these withholding periods enforced at task assignment, and the result available to anyone with a stake in the produce that comes off it. Once you have the verb, the spray-diary feature is twenty lines of UI. The audit pack is a query. The withholding calendar is a view. The buyer chain-of-custody is a permission. The compliance attestation is a template against the same record.
The trade-off is real. The unified platform won't beat the best dedicated spray diary on day one — the dedicated tool has had five years to polish that one slice. What the unified platform does is win the second slice and the third slice and the fourth, because they all share the same ledger. By year three, the unified platform is competitive on each individual slice and dominant on the connections between them. By year five, the question of whether to stitch best-of-breed isn't really being asked anymore.
The honest version
There are seven good agtech vendors who don't want this article to be persuasive. There are operators paying $400 a month across seven subscriptions who have built their workflows around the seams and reasonably don't want to start over. There are agronomists with years of muscle memory in a particular crop tool who reasonably resist switching. None of those positions are wrong; they're rational responses to where the market is today.
The case isn't that everyone should drop best-of-breed tomorrow. The case is that the structural reason best-of-breed has dominated agtech for fifteen years is not because it's the right model for agriculture. It's because nobody was willing to underwrite the longer arc of building the alternative. That's changing. The seams won't be in the customer's stack forever.
Best of breed is fine if your operation is small enough or simple enough that the seams don't matter. For everyone else, the platform that wins the next decade in agriculture is the one that treats every record as one record — and the value chain reads it together.
— The RedEarthOne team
RedEarthOne is the operating system for agriculture — one account, one ledger, five connected workspaces plus stakeholder portals, designed against the regulators and the buyer programmes from day one.