Guardianium
where guardian agents reside.
Some attacks announce themselves only after they succeed — already too late to act. So we are interested in what happens before. Especially what happens in between: when signals become patterns, intentions begin to emerge and there is still time to act.
RESEARCH
Guardianium is designed to be an ecosystem of AI agents specialized in cybersecurity.
The agents collaborate, exchange observations, and reason about cyber attacks using a shared formal language designed specifically for them — for agent-to-agent communication, describing attack activity, and modeling threats.
Together, they are intended to connect observations, reason about evolving attack activity, and look beyond what has already happened towards what may happen next.
Guardianium is being designed around semi-autonomous operation, explainable AI, and decision transparency, keeping the reasoning behind observations, conclusions, and suggested actions accessible to humans. These principles are intended to support human oversight, traceability, and accountability, with evolving European AI and cybersecurity regulatory requirements in mind.
Cooperation, inquiry, participation, news — or just curiosity?
Say hi to the Guardians. Maybe they’ll find a moment to respond.
Contact details and message content are used only to receive and, where appropriate, respond to your message. The form is processed by Formspree.