In the era of Machine Learning technology, its impact has extended across various domains, reaching countless applications. One result of this trend is the rise of learned indices, which employ Machine Learning models to enhance lookup. Among these structures, the Recursive Model Index (RMI) has proven to be particularly successful. In this work, we investigate the utilization of learned indices in two specific use cases. In Part I, we discuss the application of learned indices as hash functions, comparing them with many state-of-the-art options. The efficiency of these learned functions strongly correlates with the data distribution, and they can yield up to a 1.4x increase in probe throughput for simple datasets.