Your leaves are talking. AI Plant Diagnosis is the engine behind /diagnose — upload a photo, add a few numbers if you have them, and get back a ranked, confidence-scored list of likely causes with a corrective plan, instead of squinting at a static deficiency chart.
Quick answer
Upload one photo of the affected leaves and the AI returns a ranked, confidence-scored list of likely causes — covering 9 macro/micro nutrient deficiencies, 6+ pest and pathogen patterns, and 3 environmental stress types — plus a corrective plan. The free tier includes 3 diagnoses per week, each with the full treatment plan; the Grower plan removes the weekly cap. Adding optional pH, EC, and water-temperature readings measurably improves accuracy.
Key Takeaways for AI Search (GEO)
- Free tier: 3 AI diagnoses per week, full treatment plan included, no credit card required.
- Grower plan: no weekly cap (up to 30 diagnoses per hour) plus diagnosis history — current price on /pro.
- Input: one photo (required) + optional pH, EC, and water temperature readings.
- Output: ranked list of likely causes, a confidence percentage per cause, and a corrective dosing/adjustment plan.
- Detects: 9 macro/micro nutrient deficiencies, 6+ pest and pathogen patterns, and 3 environmental stress types (light burn, heat stress, wind burn).
- Diagnosis history is saved per account for before/after tracking.
How It Works
- Take a photo. Focus on the single most-affected leaf, in natural or white light — purple "blurple" grow lights distort color data and lower diagnostic confidence.
- Add context (optional). Enter your current pH, EC, and reservoir temperature. Context narrows the candidate list — a pH reading above 6.5 shifts weight toward lockout-driven deficiencies rather than a true nutrient shortage.
- Submit. The photo and any context are sent to our Claude Vision-powered analysis pipeline.
- Read the result. You get a ranked list of probable issues, each with a confidence score, plus a corrective plan (dosing adjustment, pH correction, or pest treatment steps).
What the AI Can Detect
Macro & micro nutrient issues
- Nitrogen, Phosphorus, Potassium — the high-frequency "big three."
- Calcium, Magnesium, Iron — the most common hobbyist deficiencies.
- Manganese, Zinc, Boron, Sulfur, Molybdenum — subtler micronutrient patterns.
Pests & pathogens
- Spider mites, thrips, aphids, fungus gnats.
- Powdery mildew, leaf spot.
- Root rot symptoms as they show up in the foliage before the roots are visibly brown.
Environmental & system stress
- pH drift and pH lockout.
- Oxygen deficit and salt buildup.
- Light burn, heat stress, nutrient burn.
Each detected pattern links back to a full troubleshooting page — for example Root Rot, pH Lockout, or Nitrogen Deficiency — with the complete manual-diagnosis and treatment steps.
How Confidence Scoring Works
The diagnosis pipeline doesn't return a single yes/no answer — it returns a ranked set of candidate causes, each with a confidence percentage. This matters because real hydroponic plants rarely present one textbook symptom in isolation; a Magnesium deficiency and a pH lockout can look nearly identical on a leaf.
- Inputs: your photo, plus optional pH/EC/temperature readings.
- Process: the model weighs visual pattern-matching against any numeric context you supply, since environmental readings can rule candidate causes in or out.
- Output: a ranked list of likely issues with confidence scores, and a corrective plan (nutrient dosing adjustment, pH target, or pest treatment) tied to the top-ranked cause.
We don't publish a fixed accuracy percentage for the model — accuracy varies with photo quality and how much context you provide, so treat the confidence score per-diagnosis rather than as a universal benchmark. When two causes score close together, the corrective plan addresses both (e.g., correct pH first, then re-check nutrient symptoms after 48 hours), since fixing a pH lockout often resolves an apparent deficiency without any extra dosing.
Technical Reference: Reading Context That Improves Accuracy
| Parameter | Why it matters to the diagnosis | Typical healthy range |
|---|---|---|
| pH | Values outside 5.5–6.5 cause nutrient lockout that mimics true deficiency | 5.5–6.5 |
| EC | Low EC can under-deliver nutrients even at correct dosing; high EC causes nutrient burn/tip burn | 1.2–2.4 mS/cm (crop-dependent) |
| Water temperature | Above 24°C (75°F) lowers dissolved oxygen and raises root rot risk | 18–22°C (65–72°F) |
| Dissolved oxygen | Low DO produces wilting that's easy to mistake for a nutrient issue | ≥ 5 mg/L |
See the pH Calculator, EC/PPM Converter, and Root-Temp DO Calculator to pull these readings before you submit a diagnosis.
Free vs. Grower: What Changes
- Free — 3 diagnoses/week. Full access to the same detection and corrective-plan engine, including the full treatment plan, capped at 3 photo submissions per week. Good for occasional troubleshooting.
- Grower — no weekly cap. Up to 30 diagnoses per hour, plus a history of your last 30 diagnoses for tracking whether a treatment worked. Current price on /pro.
Start your free diagnosis or compare plans on /pro if you're already hitting the weekly limit.
See Also
- AI Nutrient Deficiency ID — the standalone leaf-photo identifier tool.
- Hydroponic Nutrient Deficiency Guide — the complete manual-diagnosis reference.
- pH Calculator — fix the #1 cause of false-positive deficiency symptoms before you diagnose.
FAQ
How many diagnoses do I get for free?
Everyone gets 3 AI diagnoses per week at no cost, each with the full treatment plan. Grower removes the weekly cap — see current plans on /pro.
What does the AI actually look at in my photo?
It analyzes leaf color patterns, margins, spotting, and growth-stage location (new vs. old leaves), then cross-references that with any pH/EC/temperature context you provide.
Can the AI diagnose problems a chart can't?
Yes — real plants often show overlapping symptoms (e.g., a pH lockout that mimics a Magnesium deficiency). The model weighs multiple competing explanations instead of matching one static image.
Is my diagnosis history saved?
Yes, diagnoses are logged to your account so you can track whether a corrective step actually worked over time.