This thesis explores how graphical user interface (GUI) design can support the supervision of autonomous and remotely operated vessels from a shore-based control room. The work was carried out as the development of a working prototype for the Shore Control Lab, the laboratory at which the project was based. The interface had to serve two surfaces at once: an individual operator workstation and a large, shared operations display. The central concern was easy to state and hard to design for: how an operator can hold an overview of several vessels at once, recognise when one of them needs attention, and move into vessel-specific detail without losing the wider picture. The thesis is positioned as a design-oriented project rather than an operational validation study. Theory on situational awareness, cognitive load and remote supervision is used to frame the design problem and to give the design decisions a vocabulary beyond personal preference. This theoretical framing was grounded in practice through several insight activities: a mid-way wireframe meeting, recurring informal feedback from people in and around the lab, a visit to the traffic control room operated by Statens vegvesen, and a group interview with the customer centre at AtB. None of these was a controlled study, but together they kept the design connected to how people actually monitor and coordinate complex operations. The prototype itself is the main vehicle for the argument. Through it I explored fleet overview, vessel- specific views, event handling, live camera feeds, AIS map information, mission-related functions and a configurable operations display. Some of these elements are technically functional. The camera feeds and the AIS map are live, while events, weather and several status values are mocked or manually defined. The prototype is therefore best understood as a working design prototype: concrete enough to be opened, used, discussed and criticised, but deliberately short of a finished control system. A notable part of the project was the use of AI-assisted coding. Codex and ChatGPT were used to generate, revise and debug code, mostly to bridge the gap between Figma frames and a running web application. These tools made it realistic to build a working interface within the time available, but they did not make the design decisions. The interface still had to be steered by theory, feedback, visual judgement and the particular demands of the control-room setting. The main contribution is twofold: a working prototype, and a set of design requirements for fleet supervision interfaces derived from theory, insight and the act of building. The strongest parts of the prototype are the fleet overview and the deliberate separation between the workstation and the operations display. The event queue is the least mature element and remains the clearest candidate for further work. Taken as a whole, the thesis argues that a fleet supervision GUI should not simply present more data; it should structure information so that operators can see where attention is needed and coordinate their work in a shared control room.