Industrial Internet of Things is the employment of IoT concept to industrial gadgets, sensors, and other equipment, wherein these devices are connected to a network in order to collect and exchange data, thereby enhancing decision-making and automating industrial processes. However, this interconnectivity exposes Industrial IoT to countless cyber-attacks such as Spoofing, Phishing, and DDoS leading to equipment failure and safety issues in the industrial landscape. One of the most prominent and sought-after solutions for safeguarding Industrial IoT is the intrusion detection system. The mechanisms designed to hinder any unauthorized access, malicious activity, or any other security breach in real time in networks are referred to as intrusion detection system (IDS). This paper presents a novel intrusion detection approach named graph edge sentinel (GES) based on graph neural networks which classifies the network traffic as benign or attack. Two notable features of this model are the inclusion of spatial characteristics of the network being monitored and the relationship between the network flow data. The model has been evaluated on popular IoT and Industrial IoT datasets like Edge-IIoT, CICIDS 2018, and UNSW-NB15 to check its efficiency and accuracy.