Machine learning can be used for tasks that are labor intensive, tedious, or cannot be done at scale by humans. Three use cases are described: counting parasites in petri dishes to detect 70-95% accurately, classifying clinical documents to label 90% accurately, and coding doctor's notes to map to ICD-10 codes with 75-95% accuracy using rules-based approaches. Overall, initial results were promising but further work is needed to integrate models into products and ensure consistent performance.