DISCO’s artificial intelligence (DISCO AI) introduces a new approach to predictive solutions in the legal market. This white paper outlines how DISCO AI for tag predictions embodies groundbreaking legal technology due to its state-of-the-art infrastructure, unique approach to continuous learning, and tested precision and recall metrics of its predictive model. Continuous learning is disrupting the way technology-assisted review is completed. However, many of DISCO's competitors still require reviewers to follow a strict process that disrupts one's preferred workflow in order to apply predictive coding. DISCO's approach is different. We believe the legal team should drive the review and the machine should adapt to their workflow. The system should learn how to predict the lawyer's tagging behavior. DISCO’s convolutional neural network (CNN) runs on top of fastText in order to pinpoint key building blocks used to develop tag recommendations. Our AI system, in contrast to most others, understands that the phrase man bites dog is different from dog bites man. By converting words into fastText numbers, DISCO AI understands the semantic context of the documents word by word. Accuracy, recall, and enrichment are the metrics we use to measure the real-world results of our AI system. This white paper also includes best practices and recommendations on how to use DISCO AI to gain some insights for any particular case.