When Too Much Security Data Became the Risk

Vensure Employer Solutions faced unsustainable SIEM ingestion costs and analyst noise due to high-volume, low-value telemetry from rapid growth and acquisitions. The security team applied machine learning and large language models to filter out irrelevant logs (e.g., benign firewall allows), achieving an 83% reduction in firewall log ingestion, roughly $250k in annual savings, and a ~50% drop in mean time to respond while preserving threat-related events.