Harmonizing AI Agents and Human Analysts in CTI – Are CTI Agents Friends or Rivals to Junior Analysts
As Generative AI evolves from passive assistants into autonomous entities, Cyber Threat Intelligence (CTI) is entering the “AI Era.” When AI begins to “act”—initiating tasks and making decisions—a critical concern emerges: Will this autonomy deprive junior analysts of the hands-on experience needed for growth? This session explores how to hack the traditional CTI workflow to integrate agentic systems while ensuring human expertise remains the ultimate failsafe.
Engineering the Agentic Workflow: We present the design philosophy behind “AI Agents” that co-create with human analysts. Through NEC’s operational implementation, we demonstrate how these agents autonomously navigate intelligence cycles, and how humans can effectively “hack” these systems to enhance rather than replace human judgment.
Navigating the Three Barriers: In the face of CTI’s core challenges—Timeliness, Relevance, and Rationality—how do we define the boundaries of autonomous action? We explore the strategic delegation of “agency,” identifying where AI should lead and where human intervention is mission-critical to maintain high-fidelity intelligence.
Talent Development in an AI Era: When agents become indispensable team members, do they serve as digital mentors or career rivals? We share perspectives on a new model of talent development, where interacting with agentic systems accelerates the maturity of junior analysts, transforming the threat of displacement into a blueprint for professional evolution.
Takahiro Kakumaru – NEC Corporation
During my research career, I pioneered mobile network security advancements through R&D, actively participated in standardization and international collaborations (including Germany). Transitioning to cybersecurity strategy, I now lead cyber threat intelligence team, particularly in strategic intelligence. Certified in CISSP and GIAC (GCTI), I regularly share expertise at global forums such as FIRST, SANS, AVAR, CTA, and MITRE.