What occurs between the moment a commercial fishing vessel departs from shore and its return? How sustainable is its catch? Current frameworks often rely on self-reporting, which can result in errors or misrepresentation, particularly regarding bycatch. By equipping fishing vessels with on-deck cameras, leveraging edge devices for tracking and counting and transmitting predictions to the cloud, we can create a daily risk index. This index can be used to promptly alert stakeholders to suspicious activities. It integrates data from the computer vision system alongside metadata such as GPS coordinates, vessel speed and the captainʼs log. In this talk, we will outline the development and deployment of an AI- and vision-based solution for monitoring commercial fishing, including challenges encountered and lessons learned from the initial deployment.