Vectra AI, founded in 2011 and headquartered in San Jose, California, sells an AI-native security and observability platform for enterprise security teams. Its stated capabilities are on-premises and multi-cloud observability to find and reduce attack exposure, threat detection, investigation and response at machine speed, and posture improvement with tracking of risk reduction and audit readiness. The named AI feature is Attack Signal Intelligence, positioned for the agentic SOC, which uses behavioural analytics to detect attacker behaviour in humans, machines and AI agents.
The site cites 39 AI threat-detection patents, more than 90% MITRE ATT&CK coverage and results such as 80%-plus alert fidelity, 90% fewer blind spots and stopping identity attacks within 24 hours against a 292-day industry average; these are vendor claims and depend on the environment. Customers named include Blackstone, AT&T, Microsoft, Deloitte and American University. The pitch is signal quality: fewer, better-prioritised alerts for analysts already buried in them, which is the measurable thing to test in a proof of value.
Pricing is not published and evaluation starts with a demo request. In a trial, measure detection of your own red-team exercises, the volume of alerts an analyst must triage each day, and how the platform integrates with your existing SIEM, EDR and identity systems, since network detection adds most value when its findings reach the tools analysts already use.
