Fish filleting is a high value-add processing step because it converts whole fish into premium, standardised, consumer-ready products for high-value retail and foodservice channels. However, value capture is often constrained by yield losses, labour intensity, throughput bottlenecks, and seasonal labour scarcity. The pre-filleting stage, where raw fish is introduced into the processing line, remains one of the most challenging steps because fish must still be picked manually and fed into early-stage machines such as deheaders, tail-removal units, and filleting systems. This is physically demanding due to fish weights ranging from 4.0 kg to 9.0 kg and directly creates a throughput constraint. Natural variability in fish geometry, size, and colour, combined with the slippery and deformable nature of raw fish, makes high-speed automated handling of unprocessed fish particularly difficult. FIN-GRIP addresses this gap by developing an integrated flexible robotic cell with custom end-effectors for safe, efficient, and hygienic picking of fish of varied size and geometry from high-speed conveyors, together with rapid inspection and sorting using multi-modal vision, grasp-point estimation, 3D sensing, deep-learning segmentation, and AI- and data-driven software pipelines. The system will detect, localise, inspect, orient, grasp, and transfer fish under cluttered, overlapping, wet, and reflective conditions, while enabling low-latency motion planning for continuous conveyor operation. FIN-GRIP will demonstrate TRL 6 to 7 pilots in two complementary settings: a brownfield pilot upgrading an existing semi-automated flatfish filleting line at JOSMAR in Spain to an autonomous robotic feeding solution for the pre-filleting stage, and a greenfield pilot developing a representative pelagic fish picking line at PEM Automation in Ireland based on Irish end-user needs. Validation will use industrial KPIs covering throughput, handling accuracy, yield consistency, and reduced manual intervention.