The NBA solution leverages state of the art ML techniques to process large datasets which can be analyzed and provide insights for current IT issues. ML techniques were used to detect the correct resolutions from past cases. Natural Language Processing (NLP) techniques were applied to understand the communication (email messages and call transcripts) between customer and agent and extract what troubleshooting actions were taken. This data, coupled with historical part replacements and other case information serves as training data for the NBA solution. A multiclass multilabel deep learning model is used to detect the probability of each resolution to be a successful for a given case.