en.Wedoany.com Reported - AI-native cybersecurity company Armadin and agentic security operations provider TENEX.ai announced and executed a real-world attack testing operation on August 3, 2026, in Palo Alto, California, conducting security attacks against the live environment of a globally leading institution over a three-day period. Armadin's autonomous agent swarm initiated the attacks, while TENEX.ai's agent platform and Security Operations Center (SOC) received, triaged, and responded to the attack-generated telemetry in real time.
During the operation, Armadin's agent swarm generated 17 million attack actions, discovered 38 validated attack paths, and produced 238 security findings. TENEX.ai's agentic security operations platform triaged all 101,169 alerts and reconstructed the entire attack chain across 231 billion raw events. By manual estimation, completing forensic work of equivalent scale would require approximately 2,400 analyst hours, equivalent to a five-person team working for four months.
This test was not a one-off demonstration but a direct replication of the two companies' day-to-day operating models. Armadin's agent swarm operated against the institution's external perimeter, internal network, and web applications under zero-knowledge conditions, without privileged credentials, control-point whitelists, or source code access. The agent swarm first analyzed petabytes of attack-surface reconnaissance data, compiled it into a security knowledge graph, then launched 26,000 agents against the target environment, executing 1,300 attacks across more than 25,000 services. All actions passed through Armadin's control layer and were supervised by security models trained on expert human feedback. TENEX.ai's SOC received the attack activity through its detection stack, performed alert correlation and triage, and analysts used their own judgment to decide whether to escalate, rather than relying on pre-agreed response plans.
Of the 238 security findings, 98 were deemed significant. Most of these exposures stemmed from real-world deployed dependencies rather than characteristics that traditional vulnerability scanning could capture. TENEX.ai triaged all 101,169 alerts one by one and traced attacker activity across 231 billion events, reconstructing 38 attack paths. Attacker behavior accounted for only 1 in every 13,338 of the events recorded during the exercise. TENEX.ai provided evidence-based determinations for each finding, documenting attacker behavior, affected endpoints, whether existing telemetry corroborated the activity, and classifying them across 31 dimensions, including MITRE ATT&CK, OWASP Top 10, OWASP API Top 10, and CWE mappings. The platform also executed 31 documented validation checks, correlating 2,164 distinct source addresses across 89 alert rules, replacing assumptions about security gaps with documented data.
The institution involved in the operation had deployed WAF, endpoint security, and SOC coverage prior to the test, with protection levels comparable to large enterprise security programs. Armadin's agent swarm still identified and validated genuinely exploitable paths, proving that the exposures were real rather than theoretical. TENEX.ai completed triage and response in real time through the coordinated efforts of its agent platform, SOC, and human analysts. The two companies believe that autonomous offense and human-augmented defense operating against each other at machine speed are becoming the normal mode of enterprise security; continuous adversarial validation is replacing periodic assessments, and organizations still testing on an annual cadence may discover gaps in production only when it is too late.
Kevin Mandia, CEO of Armadin, stated that AI has significantly raised the ceiling of attacker capabilities, and organizations need to build defenses centered on effective autonomous security, using red AI to train blue AI. Travis Lanham, CTO of Armadin, noted that the company provides offensive AI capabilities constrained by system controls, used to train machine-speed defense systems. Eric Foster, CEO of TENEX.ai, said that the large-scale activity of AI agents and analyst investigations in this operation is precisely the operational environment organizations currently face. Venkata Koppaka, co-founder and CTO of TENEX.ai, emphasized that machine-speed attacks cannot be responded to with human-speed workflows; AI agents operate at machine speed, while human analysts remain accountable for the investigation process and outcomes.
