Medical coding is a task related to clinical billing, aiming at annotating non structured medical reports with codes describing diagnoses and treatments. This task which is generally done by coding specialists is known to be extremely difficult because of the myriad of existing codes and the long length of these documents. ICDiscovery is a coding aid application based on a machine learning process with a budget approach. The tool proposes a customizable number of codes by documents using different strategies and shows the text excerpts that allowed the model to make its predictions.