Loom Systems Ltd. is joining the growing ranks of companies applying artificial intelligence to the task of making sense out of the mass of data that overwhelms security professionals.

The company today is announcing an operational analytics platform for real-time detection and resolution that reads log files and looks for anomalies. Targeted at DevOps and information technology security professionals, and available either on-premises or in the cloud, Loom analyzes logs and semi-structured machine data without any specialized setup or configuration.

The Tel Aviv-based company says it does this by mathematically modeling the ways in which humans analyze such data, and then simulating the process via machine learning. It correlates each metric to baseline patterns in order to detect anomalies and predict future trends. The result is what the company claims is a 45 percent reduction in mean-time-to-resolution and a 93 percent reduction in log file clutter.

In this way, Loom says it can significantly reduce the need for manual pattern detection. “Ninety-nine percent of security tools give you visualization of data for analysis, but in order to extract insights you have to put a data scientist or engineer to work,” said Dror Mann, vice president of product. “That’s why security professionals spend 70 percent of their time parsing and classifying.”

Loom says its system requires no pre-processing. It can detect data types and choose the most appropriate display form, such as a gauge for temperature or a histogram for comparative values. It then determines whether a signal has shifted, as well as the type of shift that has occurred. The signal types are distinguished, and anomaly detection algorithms are tailored to fit them, the company explained. Signals are then automatically tracked in ways that complement their expected behavior.

“We operate at the intelligence layer to show you what’s changed about your situation and then identify the root cause,” Dror said. The underlying database is the Druid column-oriented distributed data store.

Additional features include real-time aggregation and correlation and access to a crowd-sourced knowledge bank with a wide range of built-in recommended resolutions. “Our team is constantly gathering information,” Dror said. “When something breaks, we ask how you’d fix it and then enter that information into our knowledge base.”

Pricing is by number of monitored instances, with unlimited data streaming. Packages are priced at $24,000, $50,000 and $100,000, with unlimited packages available. Loom said it works with any log file source.

Image courtesy of Loom Systems