AI Disclaimer
Last updated: May 2026. Read this before using AI scan features.
EcoShade uses artificial intelligence to assist with crop health analysis and treatment recommendations. AI outputs are advisory tools, not professional agronomist advice. Always consult a qualified agronomist for critical decisions affecting crop health, livestock safety, or human health.
1. How Our AI Works
EcoShade runs crop photos through three independent vision models (OpenAI, Anthropic Claude, and Google Gemini), cross-validates their diagnoses, then applies seven safety rules before anything reaches you. Production models: OpenAI gpt-4.1-mini, Anthropic claude-sonnet-4-6, Google gemini-2.5-flash-lite. When all three models agree on a diagnosis, confidence scores are higher than when they diverge.
2. Confidence Scoring
Every scan gets a composite confidence score from 0 to 100 based on four weighted factors:
- Vision Confidence (40%): Average AI model confidence in the diagnosis.
- Regional Disease Probability (25%): Historical outbreak data for your GPS region and season.
- Weather Compatibility (20%): Whether current weather conditions match the disease profile.
- Growth Stage Match (15%): Whether the detected growth stage matches your registered plant data.
Confidence thresholds:
- 60 and above (Confident): Full recommendations shown.
- 40 to 59 (Cautious): Recommendations shown with a warning. You must acknowledge before proceeding.
- Below 40 (Uncertain): No recommendations shown. Retake the photo or consult an agronomist.
3. Safety Rules
Every recommendation passes through seven hardcoded safety rules before you see it. These rules cannot be turned off in settings or overridden by AI output. Changing a rule requires a new code deployment.
Removes any recommendation containing an EU or globally prohibited active ingredient.
Caps doses at legal maximums. Warns if the AI suggested an excessive dose.
Blocks recommendations if your plant is too close to harvest (Pre-Harvest Interval).
Attaches Re-Entry Interval information to all chemical treatments.
Blocks neonicotinoids and systemic insecticides during flowering to protect pollinators.
Hides non-organic-certified products if your farm preference is organic.
Removes products banned in your country (country-specific bans vary).
4. Limitations
- AI models may misidentify plant species, diseases, or growth stages.
- Photo quality (lighting, angle, resolution) significantly affects accuracy.
- Regional disease data is updated monthly. Recent outbreaks may not be reflected immediately.
- Treatment recommendations are based on general agricultural knowledge and may not account for hyper-local conditions.
- AI cannot detect issues that are not visually apparent (root diseases, nutrient deficiencies without leaf symptoms).
5. Audit Trail
Each scan stores the models used, their individual outputs, the composite confidence score, safety rules triggered, and which recommendations were shown or blocked. You can include this history in a JSON export from Settings using Export my data.
6. Your Responsibility
By using EcoShade's AI features, you acknowledge that AI recommendations are tools to assist your decision-making, not replacements for professional agronomist consultation. You accept full responsibility for treatment decisions made based on AI outputs. EcoShade is not liable for crop losses, environmental damage, or any other consequences arising from the application of AI-generated recommendations.