The AI Agent Enterprise

Your agents work. Now decide what your enterprise becomes when it runs on them.

The engineering is done — the agents are built, hardened, and operated. The question left is the one boards, founders, and executives actually have to answer: what does a company become when it runs on agents, and how does it capture their value while containing their risk? This book takes the executive's view, and its one governing idea is a substitution that reorganizes every decision that follows: an agent is not software you buy, it is labor you rent — and the enterprise that hires, manages, governs, and holds agents accountable like workers succeeds where one that treats them as software fails.

Through ten chapters, each pairing a real enterprise failure with a decision tool you can run, you will work through:

Agents as digital labor — and why augmentation, not replacement, is the larger and more durable value Which work to delegate and, harder, what to keep human — a delegation-fit scorer where safety caps value The human–agent organization — the span of control for agents and the new "manager of agents" role no one budgeted Build, buy, or orchestrate — an answer that flips with whether a capability is your edge or your plumbing The unit economics of the agent business — cost per outcome, margin, and why the economy rewards volume and punishes pilots Governance and accountability — scoped decision rights and a named human for every decision, because "the agent decided" is not an answer Legal and compliance — executable red-line tests, because a policy an agent cannot read enforces nothing Ethics and alignment in practice — surfacing the value trade-offs your agents make so a human can own them The philosophy of agency, made operational — degrees of agency as a risk tier, and the grant-versus-oversight gap where incidents live The AI-native destination — the portfolio, the maturity ladder, and the leap from a graveyard of pilots to an operating enterprise

The governing idea, held for ten chapters: treat agents as labor, not software. Each chapter adds a research lineage (Coase and Autor to Berlin and Leveson, the RACI to the capability-maturity model), a failure catalog, an applied deep-dive, and exercises. The companion toolkit, aria7, is a set of deterministic decision models — labor economics, delegation fit, unit economics, governance, compliance, agency scoring, the roadmap — that run offline with one dependency, so every trade-off in the book is not just argued but executable.

Who it's for: executives, founders, boards, and strategists deciding whether and how to run their enterprise on agents — and the senior leaders who will own the outcome. The capstone of the series; readable on its own, richer after Books 1–6.

The AI Agent Series takes readers from first contact to professional mastery across seven volumes. This is Book 7, the last: the economics, organization, governance, and meaning of the AI-native enterprise.