The Adaptive AI SOC Agent – Moving Beyond Linear Playbooks with Cognitive Reasoning
Abstract
This paper presents an architecture for an autonomous Tier 1 security analyst designed to operate across heterogeneous SOC environments. Unlike static automation, the proposed framework utilizes a schema-agnostic abstraction layer to normalize diverse security logs into a unified investigation context. The system employs a Solver-Critic cognitive loop, where distinct agents propose and verify investigation traces against raw evidence to mitigate hallucination. This enables the autonomous triage of high-volume alerts without hardcoded integrations.
Fields of Study & Subjects
Network Security and Intrusion Detection
Information and Cyber Security
Digital and Cyber Forensics
Publication details
- Journal / source
- IEEE International Conference on AI in Cybersecurity (ICAIC)
- Pages
- 1–3
- Publication date
- 18 February 2026
- Language
- EN