Enterprise Architecture Has Four Domains. The Agentic Enterprise Needs a Fifth.

Four Domains, Thirty Years

And when your architecture review board asks which domain it belongs to, existing frameworks give you places to start an answer — a role in Zachman's Who column, a business attribute in SABSA, a stakeholder concern in TOGAF — but not a first-class place to land the full answer.

Enterprise architecture has four domains. Business. Data. Application. Technology. Security runs across all of them. For thirty years, that model held well because most of what it needed to describe was a system the enterprise owned, configured, and ran.

An agent is still a system. It is also an actor the enterprise delegates authority to — and that second description is the one existing frameworks treat as a fragment scattered across their existing structures, not a first-class concern with its own home.

The Frameworks Describe Pieces. None Owns the Whole.

Existing frameworks give architects places to describe agents, their roles, and their dependencies. TOGAF, Zachman, SABSA, and FEAF each touch a piece of the delegated-authority question. What none of them makes a first-class, explicit concern is delegated agent authority itself: who granted it, under what runtime constraints, revocable how, with what evidence left behind afterward. An agent that acts autonomously and, on occasion, reaches for a tool or scope it wasn't cleanly granted is describing a possible failure mode of delegation going wrong — not proof that no framework can represent delegation at all.

We have spent three decades modeling the systems the enterprise builds. We don't yet have a dedicated place to model the workforce it delegates to.

A Workforce With No Org Chart

The enterprise has hired a workforce it never put on an org chart — and no framework treats where that workforce lives as a first-class question.

The Fifth Domain

My proposal for that gap: Enterprise Agent Architecture. A fifth domain, alongside the other four, that makes delegated agent authority a first-class architectural concern — with explicit ownership, runtime constraints, revocation, and evidence requirements — rather than leaving it distributed as an afterthought across the other four. It has four layers:

• The workforce — which agents exist, what authority they hold, how it is delegated and revoked.

• Capability — what they are permitted to touch, enforced rather than assumed.

• The control plane — where policy meets runtime: the moment an action is allowed, denied, or held.

• Governance — who owns the agents, who audits them, and who is accountable for decisions a human did not make.

Why a dedicated domain, and not only a cross-domain viewpoint. Neither a domain nor a viewpoint assigns an owner automatically — that's an organizational decision either way. A cross-domain viewpoint can represent delegated authority, and an enterprise can assign ownership for it through existing governance structures without waiting on a new domain. What a dedicated domain adds is a named, durable place for that ownership to live and be reviewed consistently, the way Security's cross-cutting concerns eventually earned their own governance and control structures rather than staying a viewpoint over the other four. The claim here is that the dedicated home is worth having, not that the alternative is incapable of assigning an owner.

Not a Security Problem, Not an Integration Problem

This is not a security problem you can hand to the CISO, or an integration problem you can hand to the platform team. It crosses both, which is precisely why it is an architecture problem.

An Attempt to Be Early

The frameworks will catch up. They always do — late, after the enterprise has already absorbed the shift in pieces and paid for the gaps. This is an attempt to be early.

Enterprise architecture has four domains. The agentic enterprise needs a fifth. Over the coming pieces, I’ll draw it, layer by layer — starting with why TOGAF's treatment of a non-human workforce is incomplete, and what a first-class treatment adds.

The Bridge to Part 1

About This Series

Cite the Position Paper

@misc{saleme2026eaa,
  author    = {Saleme, Michael K.},
  title     = {Enterprise Agent Architecture: The Case for a Fifth Architecture Domain for the Agentic Enterprise},
  year      = {2026},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.21105314},
  url       = {https://doi.org/10.5281/zenodo.21105314}
}