BRK318 Breakout Thu 19 Nov 10:15 PST
Project Perception: How Agentic Security Works
Naama Ben Dov, Corina Feuerstein
Summary
From the recording / Microsoft AI summary
Project Perception is described as coordinated, specialised AI agents connecting data, reasoning and action across the security stack, in three layers: a data and intelligence foundation, specialist agents, and a human-controlled action layer. Agents triage and investigate SOC incidents, identify detection gaps such as Okta and AWS, and draft detections for engineer approval. They also pen test exposed web applications, validate exploitable vulnerabilities, narrow hundreds of findings to a short list, and prepare GitHub pull requests for developer approval. Caveat: based on an AI summary; products, licensing, pricing, regions and GA scope were not stated.
Session description
Microsoft's catalogue description
Stay ahead of evolving threats with Project Perception, Microsoft’s new agentic security system. See how specialized red, blue, and green team agents work together with shared security context to expose weaknesses, investigate threats, prioritize risk, and strengthen defenses. Learn how coordinated multi-agent defense helps your team move from reactive security to continuous protection, while keeping people in control of critical decisions.
Demo timestamps
Microsoft's player does not support links to a specific time, so open the recording and skip to the time shown.
- Skip to 6:38
SOC investigation with automated triage and detection coverage
Project Perception SOC investigation demo showing automated triage and detection coverage. The captions return to the demo around 33:32–34:05.