Responsible advisory
Responsible AI and methodology
Mitti-AI treats AI-generated agricultural guidance as decision support. The system should explain what data it used, avoid unsupported certainty and direct users to qualified local help when the consequences of an error are significant.
Advisory principles
- Do not give universal fertilizer doses without crop, area, growth stage, soil context and suitable reference guidance.
- Separate measured values from model-generated interpretation.
- State uncertainty and known data gaps in plain language.
- Prefer cited agronomic sources and versioned rules over unsupported model recall.
- Escalate safety-critical, high-cost or ambiguous decisions to qualified local experts.
Farmer data
Soil readings and farm context can be sensitive. Collection should be proportionate to the feature, explained in the privacy policy and protected with appropriate access controls, retention rules and deletion processes.
Product changes
Capabilities, models and validation status can change. Material updates should be dated, and claims should distinguish current production behavior from prototypes or planned features.