Enterprise Architecture

Overview

The EA Modeler is a canvas-based tool for building and maintaining your organization’s enterprise architecture. Models are created in ArchiMate 4 (the default notation for enterprise structure), with BPMN 2.0 for process/workflow behaviour and SysML v2 for systems requirements, constraints, interfaces, and verification. They are intended to be implementable — not just decorative diagrams. They connect directly to the products, technology, and operations managed elsewhere in the platform.

Key Concepts

Workroom Definitions

Open Architecture > Workrooms (/ea/workrooms) to review collaboration from the definition perspective. The page keeps all four portfolios visible, groups each Value Stream Team under its owning portfolio, and links the reusable shape to its process view and operational instances. Define or refine the underlying Value Stream Team when participants, queues, or approval triggers need to change; use Operations > Workrooms to inspect activity created from those definitions.

What You Can Do

AI Coworker Identity

Architecture views may still contain older /ea/agents links from governance and route projections. Those links are compatibility paths. They now open /platform/identity/agents, the canonical AI Coworker Identity page for principal coverage, portable identity metadata, and authorization inspection. Use the AI Workforce area for everyday coworker discovery and work assignment.

Following a Concern Across Views

Select an element on an architecture canvas and open Architecture context in the inspector. Related viewpoints lists the nearest views that explain the same element or a directly linked concern. Each link names the notation, target view, and link distance so you can see why it is being suggested.

AI routing elements can also offer Open operational evidence. This opens the AI Operations Map in Compare mode, focused on the same routing stage. The architecture remains the designed projection; the Operations Map supplies the observed evidence and calls out gaps rather than treating missing evidence as proof that the design ran.

The Why this decision works this way section deliberately shows only safe metadata. It can include decision inputs such as sensitivity class or provider eligibility, possible outcomes, the design version, implementation status, and the latest evidence time. It never exposes prompt text or protected payload values.