A diagnosis on its own is not a decision. Knowing that a crop has a particular disease tells you what is happening; it does not tell you whether to spend money this week, whether the field can wait until the weekend, or whether the answer is confident enough to act on at all.
That gap is why a diagnosis report contains more than a name. An ARCORA report returns seven things, and each one exists to answer a different question a farmer or FPO manager actually has to answer. This article goes through them in the order they matter for a decision, and — importantly — explains where the report tells you to stop trusting it.
If you want the mechanics of how the diagnosis is produced from a photograph, that is covered separately. This is about what to do once you are holding the result.
What the report contains
Seven components, each answering a distinct question:
| Component | The question it answers |
|---|---|
| Primary diagnosis | What is affecting this crop |
| Confidence score | How certain is that identification |
| Severity rating | How urgent is this |
| Treatment protocol | What specifically to do |
| Economic impact | Is acting worth the cost |
| Seasonal calendar | What comes next, and when |
| Lab referral | When to stop and get a physical test |
Most attention goes to the first line. In practice the second and fifth change more decisions.
Confidence: the number most people skip
The confidence score states how certain the identification is. It is the most consequential figure in the report and the most commonly ignored, because a name feels definitive in a way a percentage does not.
Treat it as a decision rule rather than a footnote:
High confidence. The identification is well supported by what the image showed. Proceed with the treatment protocol.
Moderate confidence. Something in the evidence was ambiguous — an early-stage symptom, a partial view, conditions that overlap with another disease. The reasonable response is usually not to spend on treatment immediately. It is to photograph again, more carefully, and see whether the confidence resolves.
Low confidence. The report is telling you it does not have enough to be sure. Acting on a low-confidence identification is how farmers spend money on the wrong chemical. This is the point at which the lab referral matters.
A system that reported only a diagnosis and never its own uncertainty would be easier to read and considerably more dangerous. A confidence score is a machine declaring the limits of what it saw.
Why the same disease can return different confidence
Confidence reflects the evidence available, not just the disease. The same infection photographed at first appearance and at full expression will not produce the same certainty, because early symptoms across many diseases look alike. This is not inconsistency — it is the system being honest about an image that genuinely contained less information.
Practical consequence: if confidence is lower than you expected, the usual cause is the photograph, not the crop. Which brings us to the most controllable variable in the whole process.
What the photograph decides
Diagnosis quality is bounded by image quality. A report can only reason about what was actually visible.
ARCORA accepts up to six images across five plant parts — leaf, stem, fruit, flower and root — in a single session. That is not a formality. A disease expressing on leaves may have its distinguishing signature on the stem, and a single leaf photograph forces the system to distinguish between conditions that look identical at that one location.
If you photograph one part, you get a diagnosis reasoned from one part. Photographing the affected area, an unaffected area for comparison, and any other part showing change is the difference between moderate and high confidence more often than anything else you control.
Severity: what changes the timeline
Severity answers urgency, and it is separate from confidence deliberately. A confidently identified minor problem and an uncertain serious one demand very different responses, and collapsing both into one number would lose that.
Severity is what determines whether this is a today problem, a this-week problem, or something to monitor. For an FPO managing many farmers across a catchment, it is also a triage instrument — it establishes which fields get attention first when there is more demand than time.
Economic impact: the line that decides whether to act at all
This is the component most diagnostic tools omit, and the one that most often changes the decision.
Knowing a crop is diseased does not settle whether to treat it. Treatment costs money — chemical, labour, time. Some interventions cost more than the loss they prevent, particularly late in a season or on a low-value crop where the remaining yield does not justify the input.
The economic line puts the potential loss next to the cost of acting. Sometimes the correct agronomic answer and the correct financial answer diverge, and a farmer deciding with a household budget in mind is entitled to see both.
For context on the scale this operates against, the losses Indian farmers absorb to crop disease are substantial — and a meaningful share is spent on treatments applied to the wrong problem, or applied when the economics never justified it.
Treatment protocol and seasonal calendar
The protocol is the specific action: what to apply, in what sequence, with what timing.
The seasonal calendar extends the same reasoning forward. A disease appearing now often indicates conditions that will produce related pressure later in the season. The calendar converts a single diagnosis into a planning input rather than a one-time fix — which is where diagnosis starts compounding into agronomy.
Lab referral: where the system says stop
The most important component is the one that declines to answer.
Some identifications cannot be made from images at all. Certain soil-borne pathogens, some deficiencies that mimic disease, and conditions requiring culture or tissue analysis are outside what any photograph can settle. When the report issues a lab referral, it is stating that the remaining question requires a physical test.
A diagnostic system that always produced an answer would be more satisfying and less trustworthy. Knowing the boundary of the method is part of the method.
How Truffaire built this
ARCORA covers 183 crops across five plant parts, returns a complete report in under two minutes, and starts at ₹33 per report. It is deployed with five farmer producer organisations in Karnataka, and has generated over 2,000 reports.
Two design decisions shaped the report format, and both came from the deployment context rather than from the model.
Speed was treated as a clinical variable, not a convenience. A diagnosis that arrives after the decision has been made is not a diagnosis. Field decisions get made on whatever information exists at the moment the farmer is standing in the field — so the target was a complete report inside the window where it can still change the action. Why speed matters this much is a longer argument in itself.
Cost had to sit below the decision it informs. A diagnostic that costs more than the treatment it might avert will not be used, however accurate. The economics only work if consulting the system is cheaper than guessing wrong once.
The report structure follows from those constraints. It is built to be read by someone standing in a field, on a phone, deciding whether to spend money today.
What we do not claim
ARCORA does not replace an agronomist, and the lab referral exists because it is not intended to. It is a first-line diagnostic that resolves the common and well-evidenced cases quickly and cheaply, and identifies the cases that need a person or a laboratory.
Nothing in the report should be read as a guarantee of yield outcome. It is decision support against a specific observation at a specific moment.
Where ARCORA fits
ARCORA is Truffaire's crop diagnosis system. A farmer photographs an affected plant part; the system returns the seven-component report described above.
It runs through FPOs rather than direct to individual farmers, for a specific reason: the FPO already has the relationship, the trust and the distribution. What an FPO is and why that structure matters covers this, and how the five FPOs in the network accumulate shared diagnostic data covers what the aggregate becomes over seasons.
The data generated belongs to the network that generated it. That is a stated commitment, not a feature.
Frequently asked questions
What should I do if confidence is low?
Photograph again before spending anything. Capture the affected area, an unaffected area of the same plant, and any other plant part showing change. Low confidence usually reflects limited visual evidence rather than an unusual disease. If it stays low, the lab referral is the correct next step.
Can I photograph just one leaf?
You can, and you will often get a usable answer. But single-part images are the most common cause of moderate rather than high confidence, because many diseases are indistinguishable on one plant part in early stages. Up to six images across parts is available for exactly this reason.
Does a high severity rating mean I will lose the crop?
No. Severity describes urgency of response, not an outcome forecast. A high rating means the window for effective action is short — which is a reason to act, not a prediction that action will fail.
Why does the report tell me the cost of treating?
Because the agronomically correct answer and the financially correct answer are not always the same, particularly late in a season. A farmer deciding against a real budget should see both, rather than being told only what is technically optimal.
What if the report recommends a lab test I cannot access easily?
Then the honest position is that the identification is unresolved. Treating on an unresolved diagnosis is a gamble with input costs. Where a lab is genuinely inaccessible, the FPO is usually the right route — several maintain relationships with agricultural universities and KVKs for exactly this.
Where to start
Read the report in this order: confidence first, because it establishes whether the rest is actionable. Severity second, because it sets the timeline. Economic impact third, because it decides whether acting is worth it. Then the protocol.
If confidence is low, the next step is a better photograph, not a purchase. If the report issues a lab referral, that is the system being precise about its own limits — which is the part that makes the rest of it trustworthy.
If you work with an FPO and want to understand how ARCORA is deployed, get in touch, or see where it sits in Truffaire's systems.