Streem.ai, is an unsupervised, real-time multivariate anomaly detection and pattern recognition tool which integrates domain knowledge into a robust classification of discovered faults in time-series data. Streem.ai digests large complex data, visualises patterns, and notifies engineers in case of abnormal or new behavior in real-time. Stream.ai’s product realizes that unsupervised machine learning (i.e. without labeled data) allows us to do anomaly detection analysis on all sensor data at scale as the same algorithms can be applied across industry/sensor types without the need to understand the specific domain and set expensive labels generated per client. Using this technology we can also identify outliers.