How Kisan Salahkaar Works

This page explains what inputs the app uses, which systems support those flows, and where users should be careful about verification.

It is written to make the product easier to evaluate by users, reviewers, search systems, and LLM-based agents.

Inputs the product uses

Different features rely on different inputs. Crop recommendations use user-provided location, soil type, and weather pattern details. Pest and disease diagnosis uses an uploaded plant image. Complete recommendations use an uploaded soil report plus location context.

Generation and model layer

The app uses Google Gemini through Genkit for text generation and reasoning. That model layer produces the generated agricultural guidance shown in the app.

Outputs are generated from the user inputs supplied during the session rather than from a public article database maintained on the site.

Supporting services

Location autocomplete uses OpenStreetMap Nominatim. Browser geolocation may be used when the user asks the app to detect their current location. Product analytics use Firebase Analytics when supported in the browser.

• - OpenStreetMap Nominatim for location lookup.
• - Browser geolocation for user-requested device location.
• - Firebase Analytics for product telemetry.

Current limitations

Weather and market outputs in the current public app are not backed by a dedicated live official weather feed or mandi pricing feed inside the product. They are AI-generated guidance tied to the location the user provides.

Because of that, the app should not be treated as the final authority for spraying windows, irrigation timing, harvest timing, or price-sensitive selling decisions.

• - Verify weather-sensitive decisions with a trusted live weather source.
• - Verify pricing with the relevant mandi, market, or trading channel.
• - Verify pesticide or fertilizer actions with a qualified agronomy source before field use.