Naga Krishna Reddy Muppidi

Talk title: Governing Agentic AI in Regulated Financial Infrastructure: CostAgent, SecReviewAgent, and the Path to Auditable Autonomy

Talk category: AI governance, agentic AI, cloud infrastructure, cybersecurity, enterprise AI, regulated financial infrastructure

Talk abstract: As enterprises adopt agentic AI for cloud operations, infrastructure governance, and security review, regulated industries face a difficult balance: gaining automation without losing compliance control. Financial infrastructure environments require explainable decisions, approval boundaries, evidence retention, audit trails, and clear ownership for every operational change.

This talk introduces a practical governance framework for agentic AI in regulated financial infrastructure. It uses CostAgent and SecReviewAgent as two applied examples. CostAgent illustrates how agentic AI can assist with cloud cost optimization under bounded action spaces, policy constraints, and approval gates. SecReviewAgent illustrates how AI-assisted review can support infrastructure-as-code security analysis, architecture memory, evidence capture, and reviewer-oriented explanations.

The session covers how to define agent authority, separate recommendation from execution, preserve human approval for high-risk actions, capture evidence for audit review, and measure agentic AI systems using metrics such as cost savings, review latency, policy violations avoided, unsupported claims, and audit readiness. The goal is to show a path toward auditable autonomy: agentic AI systems that improve operational efficiency while remaining governable in financial-services environments.

Audience takeaways: - Understand why agentic AI requires stronger governance in regulated financial infrastructure. - Learn how CostAgent and SecReviewAgent illustrate bounded autonomy for cost and security workflows. - Learn how to design approval gates, policy boundaries, evidence retention, and audit trails for AI agents. - Understand how to evaluate agentic AI systems using operational and compliance-focused metrics. - Leave with a practical adoption model for auditable AI-assisted infrastructure automation.