The effective management of fisheries species requires the conservation of the resources they need throughout their lives. Understanding the diet of predatory fish provides vaulable insight into their resource needs, yet diet studies face a significant challenge: the rapid rate of prey digestion. Within half to two hours, fish prey in the guts of warm-water predators are no longer identifiable from external morphology, and diet studies commonly report up to 80% of fish prey as 'unidentified fish'. However, many species can be identified from their otoliths which remain intact for many hours. We are developing a web-based tool using machine learning, to compare user-uploaded otolith images to a reference database, currently >2800 images from >60 spp from the northen Gulf of Mexico. Users are presented with the closest matches to their unknown otolith, summary statistics allowing evaluation of how reliable the match is at the species, Genus, Family, and Order levels, and can browse images from matching classes. Users have the option to input the weight of their otolith, and the tool will provide an estimate of the size of the fish it came from, based on otolith weight-fish size regressions we are developing for each species. The tool is performing successfully, with most errors due to limited sample sizes for species or size classes with similar shaped otoliths (e.g. Threadfin shad and Scaled herring). In addition to internal cross-validation of the tool, we will test it on otoliths found in the stomachs of spotted seatrout from Mobile Bay. The tool will greatly enhance the resolution and efficiency of diet studies, providing more information from every fish sampled. We are inviting potential users to test the tool and provide feedback, and to help expand our image library with contributions from other species and regions.