Bodungen on broken governance, agentic AI and MindStone

Clint Bodungen, director of AI/ML engineering at Arcovo and founder of ThreatGen, discussed failures in cybersecurity governance, the role of people in risk, the open-source MindStone Agent and a recent ransomware response in an interview at the ICS Cybersecurity Conference in Nashville.

Bodungen said governance frameworks often become checklists that do not match industrial operational realities. He noted practitioners must reconcile policy requirements with production constraints, and that mismatch creates gaps for defenders to manage during incidents.

He identified human factors as the largest source of vulnerability. Training gaps, incentive structures and organizational culture were cited as common causes that allow attackers to exploit systems faster than technical fixes can be applied. He argued governance should connect to daily workflows to be useful to engineers and incident responders.

Bodungen publicly described the MindStone Agent for the first time. The project is open-source and aims to give AI assistants persistent memory, a consistent identity and continuity across tasks and sessions. The design intends for an assistant to retain prior context, keep a defined role over time and carry forward multi-step work after interruptions so processes do not lose state between interactions.

He provided a technical account of a ransomware incident where multiple autonomous agents coordinated parts of the response. The agents performed forensic analysis, handled containment, executed recovery steps and migrated infrastructure with limited human oversight. The system’s ability to preserve context and continue tasks across stages supported the agents in completing multi-step operations while reducing manual steps for the security team.

Bodungen’s background includes founding ThreatGen and leading AI and machine learning efforts at Arcovo. The interview reviewed changes in cybersecurity practices over the past two decades, examples of where governance struggles to help defenders, the MindStone Agent’s goals for continuity in assistant applications, and an operational example of agent-driven incident work.