| Agent |
An AI-powered entity that operates with a degree of autonomy within the enterprise — making decisions, executing tasks, and collaborating with humans and other agents within defined authority boundaries. Distinguished from a "tool" by its capacity for autonomous action. |
| Agent Auditor |
A role responsible for independent assessment of agent compliance, performance, and governance adherence. |
| Agent Owner |
The human or team accountable for a specific agent's configuration, performance, and fitness for purpose. |
| Agent Platform |
The technology infrastructure, tooling, and operations that enable enterprise agents to operate. Managed within the Technology & Data domain. |
| Agentic Maturity Profile |
A structured assessment of a domain's readiness for agent participation, showing the current and target human/agent split across activities. |
| Authority Boundary |
The explicitly defined limits of an agent's decision-making and action-taking power. Agents must not act outside their authority boundary. |
| Authority Grantor |
The human who approved an agent's autonomy level and authority boundaries. Accountable for the appropriateness of the authority granted. |
| Autonomous Agent |
An agent type that executes defined work independently within bounded authority, without requiring human involvement for each instance. See Workforce Model. |
| Autonomy Level |
A classification (A0–A4) that defines how much independence an agent has. Ranges from A0 (Inert — information only) to A4 (Autonomous — full authority within scope). |
| Blended Workforce |
An organisational workforce comprising both human workers and AI agents, operating as an integrated team with defined roles, responsibilities, and governance. |
| Capability Composition Matrix |
A ZORBA assessment tool for determining the appropriate workforce composition (human, agent, or blended) for each capability. |
| Chain of Accountability |
ZORBA's model for resolving the accountability question when agents act. Maps accountability through Agent Owner, Authority Grantor, Process Owner, Domain Owner, and Governance Function. |
| Chief Agent Officer |
An executive role responsible for enterprise-wide agent strategy, governance, and risk management. |
| Co-Pilot Agent |
An agent type that works alongside a specific human, augmenting their capabilities while the human retains decision authority. See Workforce Model. |
| Composition Profile |
A classification of the human/agent workforce mix for a function: Human-Essential (H), Human-Led (H+a), Blended (H=A), Agent-Led (h+A), or Agent-Essential (A). |
| Decision Authority Matrix (DAM) |
A governance artefact that defines, for every significant decision type, whether a human or agent has authority to make it, and under what conditions. |
| Domain Agent Steward |
A role responsible for agent governance within a specific enterprise domain. |
| Escalation |
The mechanism by which an agent transfers control to a human (or higher-authority agent) when encountering situations outside its authority or capability. A design feature, not a failure mode. |
| Escalation Packet |
The structured bundle of information an agent must provide when escalating: context, actions taken, assessment, recommendations, and state. |
| Execution Orchestrator |
An agentic pattern in which an agent manages end-to-end process execution, routing work to appropriate human and agent performers. |
| Governance Gradient |
The ZORBA concept that governance intensity varies across layers — highest at Strategy, evolving to continuous monitoring at Work. |
| Governance Maturity Level |
A classification (G1–G5) of how mature an organisation's agent governance is, from Ad Hoc to Optimising. |
| Human Override |
The foundational governance principle that any human with appropriate authority can stop, redirect, or take over agent activity at any time. Non-negotiable in ZORBA. |
| Orchestrator Agent |
An agent type that coordinates and manages workflows involving multiple humans and/or agents. See Workforce Model. |
| Specialist Agent |
An agent type with deep expertise in a narrow domain, invoked for specific tasks. See Workforce Model. |
| Trust Calibration |
The ongoing process of adjusting an agent's autonomy level based on observed performance, trust signals, and changing context. |
| Trust Signal |
An observable indicator (positive or negative) that informs trust calibration decisions. Examples: decision quality, escalation appropriateness, error rates. |
| Workforce Composition |
The designed blend of human and agent capability for a given function, process, or activity. An explicit architectural decision in ZORBA. |
| Workforce Composition Audit |
An assessment of whether the actual human/agent mix matches the designed composition, and whether the designed composition remains appropriate. |