The growing integration of Internet of Things (IoT) sensor networks and artificial intelligence (AI) is driving a digital revolution in precision agriculture, and at the same time, presents new issues in maintaining the quality and reliability of sensor data through Enterprise Information Systems (EIS). The current paper outlines a proof-of-concept case study of an AI agent-based framework of automated sensor error detection and data-quality management in agricultural IoT platforms. The system is implemented as a low-code application that uses modular AI agents to coordinate the ingestion of data, exploratory analysis, anomaly detection, and structured reporting of multiple heterogeneous channels of environmental sensors. The prototype was tested on real-world data collected by agricultural monitoring stations, where agents produced daily and historical summary reports, identified both “hard” errors (e.g., missing, zero, or physically implausible values) and also the subtler problems (e.g., calibration drift or emergent risks), and provided recommendations in the form of actions to maintain sensors. Although the existing prototype is still at the pilot level and uses basic statistical thresholds, the modular architecture demonstrates its feasibility, transparency, and accessibility to non-expert users. The article highlights the possibilities of integrating low-code automation, AI agents, and large language models (LLMs) to facilitate early warning systems and data-quality management in agricultural IoT systems. However, it also recognizes the need for bigger scale validation, benchmarking, and integrating with EIS in future studies.
Chojka et al. (Thu,) studied this question.