SMARTSAFRA: INTELLIGENT CHATBOT FOR AGRICULTURAL DATA COLLECTION AND MONITORING

Abstract

The lack of accessible tools for crop management and the absence of structured data allowing for a consolidated view of production indicators by community represent two facets of the same problem faced by small-scale Brazilian farmers. This study presents SmartSafra, a chatbot designed for small-scale corn and soybean farmers that enables the structured collection of production data and the generation of financial indicators, operating entirely via WhatsApp. The system was developed using a conversational state machine—without relying on generative artificial intelligence—and is complemented by a web dashboard for farmers to visualize their data and an administrative dashboard providing managers (such as municipal agriculture departments) with a consolidated view of indicators by community. Validation was conducted through functional black-box testing based on the IEEE 829 standard, organized into three scenarios: the complete usage workflow, handling of invalid inputs, and data scraping/market quotes. The seven financial indicators calculated by the system—including yield per hectare, gross margin, and break-even point—were verified via desk checking. All tests across the three scenarios passed, with the system successfully collecting crop data, calculating indicators, and returning performance comparisons to the farmer. A limitation of the study is that validation was conducted exclusively using simulated data in a controlled environment, without exposure to actual farmers; this step has been designated for future work. The study concludes that the adopted approach is a viable way to provide small-scale farmers with an accessible crop management tool—requiring no app installations or specialized hardware—while simultaneously enabling managers to access an aggregated view of production indicators by community.