About the book
AI agents are no longer just answering questions. They are beginning to act.
They access systems, invoke tools, make decisions and execute transactions with growing autonomy. That creates a fundamental challenge for every organization: how do you give an AI agent enough authority to be useful without giving it more authority than it should have?
Bounding Agency is about answering that question: letting software act on the enterprise’s behalf, within limits someone accountable has set, and with evidence of what it did.
Drawing on established identity and access management principles while facing the realities of agentic AI, the book shows how to establish an agent’s identity, authenticate it, define and constrain its authority, govern delegated access, keep a human accountable and manage the agent from creation through retirement.
What you’ll learn
- Why AI agents create a different identity challenge from humans, machines and workloads
- Authentication and credential strategies for AI agents
- Authorization, delegation and least privilege when agents act for people and for other agents
- How to bound an agent’s authority as context changes
- Governance, accountability, monitoring and auditability
- Which existing IAM capabilities to keep, which to adapt and what to build new
- How to assess your organization’s readiness and build a 12–24 month roadmap
- A 90-day action plan for the practitioner leading this work
Who it’s for
- IAM and cybersecurity practitioners
- Security and enterprise architects
- Technology and product leaders
- AI and platform engineering teams
- Risk, compliance and audit professionals
- Governance and executive leaders
A running worked example, the fictional CrystalSoft Energy, follows two agents through every chapter.
What’s inside
Seventeen chapters in five parts, framed by an opening and a closing chapter, with the toolkit at a glance, a glossary and the full reference list at the back.
When Identity Starts to Act
What kind of actor an agent is, the difference between identity, agency and authority, and why today’s IAM is necessary but not sufficient.
- 1From Human Identity to Agent Identity
- 2Identity, Agency and Authority
- 3Why Today’s IAM Is Necessary—but Not Sufficient
The Agentic IAM Core
Identity, authentication, authorization, delegated authority, and privilege and lifecycle.
- 4Establishing Agent Identity
- 5Authenticating the Agentic Enterprise
- 6Authorization in a World of Autonomous Action
- 7Delegated Authority and the Chain of Trust
- 8Privilege, Lifecycle and Access Governance
IAM 360°
Architecture, risk, governance and regulation, and the cross-functional work IAM cannot do alone.
- 9Architecture and the Technology Ecosystem
- 10Identity Risk and the Agentic Threat Surface
- 11Governance, Regulation and Accountability
- 12IAM Cannot Do This Alone
From IAM Today to Agentic IAM
An operating model, an assessment and a roadmap.
- 13The Agentic IAM Operating Model
- 14The Agentic IAM 360° Assessment
- 15From Assessment to Roadmap
The Practitioner Playbook
Your first 90 days, and the toolkit.
- 16Your First 90 Days
- 17The Agentic IAM Practitioner Toolkit
The free practitioner toolkit
Every tool in the book, as an editable Excel workbook and a PDF, free and with no sign-up. Start with the core four: see what exists, assess it, decide the response and act.
AI Agent Inventory
What is acting in the environment, what can it reach and do, who owns it, and under what authority?
Excel 18 KBPDF 64 KBIAM 360° Readiness Assessment, KEEP → ADAPT → BUILD (KAB) Worksheet and Maturity Assessment
Where do we stand across the eight spokes, and what do we keep, adapt or build? One workbook, one evidence base, with the heatmap. The Maturity Assessment inside it is situational: use it when leadership needs a descriptive view of progress.
Excel 28 KBPDF 74 KB90-Day Action Plan
What do you do in your first 90 days, and what escalates immediately?
Excel 18 KBWord 16 KBPDF 133 KBAbout the author
Rahul Bhardwaj is a technology and product leader with more than 15 years of experience across identity and access management, cybersecurity, AI and emerging technology. His career spans payments and financial services, cloud service providers, banking, early-stage technology ventures and academia, giving him a perspective that bridges large-scale enterprise environments, startup innovation and practitioner education.
His work has included building and scaling products across digital identity, authentication, access, security, risk and trust, as well as developing AI-powered products and solutions. His current focus is the emerging intersection of AI and IAM—particularly how organizations establish identity, bound authority, govern access and maintain accountability as AI agents become active participants in enterprise environments.