4. Terms and Definitions

(Normative)

For the purposes of this specification, the following terms and definitions apply. Terms used in normative clauses that are defined in this clause are rendered in bold at first use in each clause.

4.1 Activity — The fundamental unit of work in a process model, carrying governed attributes per Clause 8.

4.2 Agent species — An empirically observed cluster of agent deployment patterns sharing governance configuration characteristics.

4.3 BPM — Business Process Management — the discipline codified in the ABPMP BPM CBOK.

4.4 BPM/Agent Stack — This specification — execution governance for AI agent architectures.

4.5 BPMN 2.0 — Business Process Model and Notation — OMG standard for process modeling.

4.6 Constitutional AI — The AI model’s training-time values and character. The foundation beneath the Intent Stack and BPM/Agent Stack.

4.7 Context — The information available to an LLM at inference time. The vehicle for intent communication, not intent itself. (Clause 11)

4.8 Dark factory — An agent deployment pattern with near-zero human involvement between specification input and evaluation-passing output.

4.9 DMN 1.0 — Decision Model and Notation — OMG standard for decision modeling.

4.10 Execution — One of three execution governance concerns claimed by this specification. How authorized work is performed within the full governance context, with process instantiation, live state, and audit trail. Distinguished from Intent Stack L1 (Runtime Alignment), which assesses whether execution outcomes align with governing intent.

4.11 Five Intent Primitives — Purpose, Direction, Boundaries, End State, Key Tasks — the irreducible governance content at every governance interface. NOT layers. (Intent Stack Clause 5)

4.12 Gateway — A process element controlling branching and merging of execution paths, per BPMN 2.0.

4.13 Governed activity — An activity carrying the full set of BPM/Agent Stack governance attributes (21 attributes, Clause 8).

4.14 Governance configuration — The specific arrangement of governance elements — Intent Primitives, human position, quality gates, trust calibration — at a deployment’s delegation interfaces.

4.15 Holdout principle — Maintaining acceptance criteria that the implementing agent never sees, preventing optimization for evaluation rather than genuine compliance.

4.16 Information scent — Expert-detectable signals in published material indicating depth behind what is shown (Pirolli and Card, 1999).

4.17 Integration — One of three execution governance concerns claimed by this specification. How governed agents connect to external systems, with typed system attributes, governed tool/API/MCP access, and governance context propagation. Distinguished from any general use of the term “integration” in software engineering.

4.18 Intent — The content of governance at a delegation interface. Relational, processual, normative. Decomposed into five primitives. Not a synonym for “what the user wants.” (Clause 11)

4.19 Intent Stack — The companion specification — a four-layer reference model for governing AI agent behavior within organizations (intentstack.org/spec/2026-04-01).

4.20 Key Tasks — The fifth Intent Primitive — defines authorized scope of work. The primary stitching primitive connecting governance context to execution structure. (Clause 7)

4.21 Memory — The mechanism for information persistence across sessions. Not intent itself. (Clause 11)

4.22 Monotonic accumulation — The structural property of Boundaries: constraints can only be added, never removed, as they propagate through governance interfaces.

4.23 Orchestration — One of three execution governance concerns claimed by this specification. How multiple agents are coordinated to execute specifications, including delegation level translation, swimlane assignment, and knowledge provisioning. Distinguished from Intent Stack layers, which address governance context rather than execution coordination.

4.24 Orthogonality — The design property of the Intent Stack and BPM/Agent Stack: neither duplicates nor complicates the other. To be preserved when extending either specification.

4.25 Four governance context layers — The Intent Stack’s four governance concerns in vertical composition: L4 Intent Discovery, L3 Intent Formalization, L2 Specification, L1 Runtime Alignment, with Constitutional AI substrate beneath.

4.26 Three execution governance concerns — This specification’s three concerns: Orchestration, Integration, Execution. Together with the four governance context layers, these constitute the seven governance concerns of the complete architecture.

4.27 Stitching mechanism — The structural connection between the Intent Stack and BPM/Agent Stack, primarily at the Key Tasks primitive operationalized through Intent Stack L2 (Specification), with evidence flowing back through Intent Stack L1 (Runtime Alignment).

4.28 Three-layer architecture — Constitutional AI (substrate) + Intent Stack (governance context) + BPM/Agent Stack (execution structure). All three necessary. None substitutes.