AI is changing the economics of cyberattacks. It can search larger attack surfaces, generate more activity than analysts can triage, and keep operating long after the SOC goes home. Defending against that shift takes more than another copilot or a faster alert queue. It requires a security system that can continuously perceive, reason, and act—without taking people out of command.

Maarten begins with why the tempo of cyber defense is changing. We will then examine the Microsoft Security platform many organizations already operate, connecting signals and controls across Defender, Sentinel, Entra, and Purview to the visibility, context, and actuators this new approach requires.

Join Maarten, a 20 year+ Microsoft MVP and go inside Microsoft Project Perception: its shared security context, multi-model harness, and specialized red, blue, and green agents. We will trace how these agents discover attack paths, investigate meaningful risk, and turn findings into corrective action as one continuous loop.

Finally, Maarten will map this architecture to the attacks defenders now face—and explore a hopeful future in which autonomous defense amplifies security teams while humans remain firmly in control.

After this session, attendees will be able to:

• Explain why machine-speed attacks require a continuous perceive-reason-act cycle instead of simply producing more alerts.
• Map Defender, Sentinel, Entra, and Purview capabilities to the signals, security context, and enforcement foundation behind Project Perception.
• Describe how red, blue, and green agents collaborate across attack-path discovery, investigation, and remediation—and identify where human control belongs.

Target audience: Microsoft security practitioners, SOC analysts, security architects, and Defender/Sentinel engineers.
Level: 200
Preferred duration: 45–60 minutes.
Format: Strategic-to-technical conference session; no special technical requirements.