Case Study: Mitiga achieves 4x faster detection and 95% less false-positive noise with Splunk AI Toolkit and Cisco Deep Time Series Model

Mitiga reduces collection failure detection time by 4x with Splunk

Mitiga, a security platform providing cloud protection, faced a significant challenge in ensuring reliable data collection for its customers. They needed to quickly detect when log collection pipelines failed, as a volume drop could signal a critical blind spot during an attack. Prior to using Splunk's AI Toolkit and the Cisco Deep Time Series Model (CDTSM), confirming these failures could take up to 24 hours.

By implementing Splunk's solution, Mitiga used the CDTSM for anomaly detection to monitor over 10,000 collection jobs. The zero-shot foundation model forecasts expected volume and flags drops outside the normal range. This allowed Mitiga to reduce detection time for broken collections from a day to just a few hours, achieving a 4x improvement in mean time to detect and a 95% reduction in false positive alerts.

Mitiga