This document discusses anomaly detection in streaming data using Hierarchical Temporal Memory (HTM). It describes how HTM can be used to build a real-time anomaly detection system that continuously learns and predicts patterns in streaming data. It also introduces the Numenta Anomaly Benchmark (NAB), an open benchmark for evaluating streaming anomaly detection algorithms that contains labeled real-world data streams and a scoring system that rewards early detection of anomalies. Detection results on sample data streams are shown for HTM and two other algorithms to demonstrate HTM's ability to detect anomalies earlier. The document promotes the NAB competition for submitting new algorithms or data to be tested on the benchmark.