Cato Networks has introduced a new security capability that uses autonomous agents to predict likely attack paths and automatically personalize protections for each customer environment before AI-assisted attacks can advance. The company, which operates a converged network and security cloud, unveiled Cato Agentic Threat Prevention at Black Hat USA 2026, marking a shift from reactive detection to predictive defense against frontier AI adversaries. The capability aims to anticipate how attackers will move through networks and deploy tailored safeguards before damage occurs.
The new system transforms Cato's unified data lake into customer-specific protection by combining platform telemetry from security and networking operations, customer-specific activity patterns, and threat intelligence. According to Cato Networks, this integration models risk across users, applications, traffic patterns, assets, and exposures, allowing the capability to forecast how agentic attackers might chain techniques together, exploit security gaps, and circumvent existing controls. The system then generates protections customized to each customer's unique environment rather than depending on universal detections. These safeguards are enforced through every Cato Point of Presence globally, with no service chaining or enforcement gaps.
"Yesterday's security cannot stop today's AI-powered attacks," said Shlomo Kramer, co-founder and CEO of Cato Networks. "Prediction is the new prevention." He explained that staying ahead of AI-powered attacks requires anticipating where they'll move next and deploying protections before damage occurs. The company states that frontier AI has created the need for agentic defense across two fronts: protection and prevention. While Cato Agentic Threat Prevention operates on the prevention side by predicting likely attack paths and creating customized protections before attackers advance, the company's Agentic CVE Mitigation handles the protection side by autonomously assessing and applying safeguards for newly disclosed vulnerabilities.
The report frames this development around a fundamental shift in the threat landscape. Frank Dickson, group vice president at IDC, notes that AI-assisted attacks are revealing the limitations of static security controls and manual response processes. An entire attack can now require less time than it takes to investigate a single alert, according to the report. As attackers adapt more quickly and customize campaigns to individual environments, enterprises will need prevention approaches that leverage shared context, automation, and cloud-scale enforcement to reduce exposure before compromise happens. The capability is built on Cato's cloud-native platform, which combines network and security telemetry into a unified view of each environment and enables AI agents to anticipate threats and continuously adapt protections.
Cato Networks positions this launch as leveling the playing field between defenders and AI-powered adversaries. The company argues that conventional security approaches can't match the speed and adaptability of modern attacks, making prediction-based prevention essential for stopping breaches before they progress through multiple stages. For security teams struggling to keep pace with increasingly sophisticated and rapid-fire campaigns, the shift from alert investigation to automated path prediction represents a fundamental rethinking of when and how defenses engage. Organizations facing AI-enabled threats may find that their response timelines no longer align with attack velocity, forcing a choice between automation and perpetual disadvantage.