Agentic Auditing

AI agents are moving from isolated assistance into workflows that can plan, retrieve evidence, call tools, maintain state, and act within institutional processes. Agentic Auditing examines what this shift means for public-sector control, auditing, and management.

Marcelo Campos Antunes develops a governance-centered approach to agentic systems, addressing autonomy boundaries, identity and authorization, evidence provenance, human oversight, reproducibility, cybersecurity, monitoring, contestability, legal-basis versioning, and institutional accountability.

Across 28 chapters, the book presents practical frameworks, including maturity models, implementation programs, operational envelopes, checkpoints, and auditing architectures, while distinguishing legal rules, official guidance, technical references, and original methodological proposals.

Designed for auditors, controllers, public managers, legal professionals, and technology specialists, this book offers a structured approach to evaluating where AI agents can add value-and where stronger safeguards, human review, or deliberately limited automation are the more responsible choices.