Enterprise Intelligence Architecture — Canonical Definitions
The authoritative enterprise vocabulary for representation, memory, coordination, consequence and learning
Document type: Arqua Canonical Vocabulary
Status: Published canonical reference
Architecture version: Arqua Canonical Architecture v1.0
Vocabulary scope: Enterprise-wide
Last reviewed: 11 July 2026
Canonical definitions may be clarified without changing their architectural meaning. Any material change to scope, ownership, parentage or relationship requires a versioned architecture decision.
This page defines the enterprise-wide vocabulary used across the Arqua architecture corpus.
These definitions establish the common meaning of the constructs that connect operational reality, enterprise representation, institutional memory, runtime context, governed coordination, execution and learning.
Domain-specific definition pages inherit from this vocabulary and may add specialised terms, but must not redefine the enterprise-level concepts without an explicit versioned architecture decision.
Standard fields for each definition: Term, Category, Canonical definition, Purpose, What it is not, Inputs, Outputs, Where it sits, Canonical parent, Related constructs, Canonical statement.
Definitions
Enterprise Intelligence Architecture
Category: Enterprise reference architecture
Canonical definition: The architecture through which an institution represents operational reality, forms governable understanding, preserves institutional continuity, assembles runtime context, coordinates action, governs consequence and learns from outcomes.
Purpose: To preserve coherent institutional understanding and self-governance as the enterprise changes.
What it is not: Not a data platform, AI platform, metadata catalogue, ontology, operating model or control system in isolation.
Where it sits: The whole Arqua architecture.
Canonical parent: Arqua Canonical Architecture
Canonical statement: Enterprise Intelligence Architecture governs how an institution understands, decides, acts and learns without losing continuity.
Operational Reality
Category: Foundational condition
Canonical definition: The enterprise as it actually exists and operates, independently of any single model, database, document, ontology, report or institutional belief about it.
Purpose: To establish that institutional representations are always representations of reality, not reality itself.
What it is not: Not a system of record, digital twin, graph, data model or accepted narrative.
Outputs: Observable events, states, behaviours, relationships and evidence.
Canonical statement: Reality precedes representation.
Observed Evidence
Category: Evidential input
Canonical definition: The records, events, measurements, documents, assertions, behaviours, traces and observations through which aspects of operational reality become available for institutional interpretation.
Purpose: To provide traceable grounds for forming, validating or challenging institutional representations.
What it is not: Not automatically accepted truth and not merely raw data.
Inputs: Operational events, records, documents, communications, sensor observations, system behaviour and human assertions.
Outputs: Evidence available for interpretation and representation.
Canonical statement: Evidence makes reality observable but does not make meaning authoritative.
System Model Foundation
Category: Representational substrate
Canonical definition: The governed representational substrate through which boundaries, subjects, identities, relationships, states, events, contexts, claims, evidence, representation types and projections become coherently expressible across the enterprise.
Purpose: To make operational reality contractably representable.
What it is not: Not a single canonical database schema, ontology, knowledge graph or industry data model.
Inputs: Observed evidence, domain knowledge, operational structures and institutional distinctions.
Outputs: Coherent representational structures available for semantic acceptance and projection.
Canonical statement: Semantic contracts require contractable representations; contractable representations require a System Model Foundation.
Enterprise Representation Intelligence
Category: Continuously operating enterprise capability
Canonical definition: The continuously operating capability that discovers, interprets, reconciles, validates, governs and evolves the institution’s representations of operational reality.
Purpose: To keep institutional understanding aligned with changing operational reality.
What it is not: Not metadata enrichment, ontology management, knowledge-graph storage, AI inference or data quality in isolation.
Inputs: Observed evidence, existing representations, semantic policies, behavioural observations, outcomes and institutional decisions.
Outputs: Proposed representations, validated representations, representation challenges, accepted revisions and memory updates.
Where it sits: Between observed evidence and accepted institutional representation.
Canonical parent: Enterprise Intelligence Architecture
Related constructs: System Model Foundation, Intelligent Semantic Architecture, Institutional Memory, Representation Graph and Behaviour Validation.
Canonical statement: Enterprise Representation Intelligence continuously maintains institutional understanding through governed representation.
Accepted Institutional Representation
Category: Governed representation state
Canonical definition: A representation of operational reality that has passed the required institutional processes of interpretation, evidence assessment, semantic governance, authority and acceptance for a declared scope and purpose.
Purpose: To distinguish institutionally relied-upon meaning from raw evidence, inference, opinion and unaccepted model output.
What it is not: Not immutable truth, raw metadata, an AI-generated claim or an ontology statement merely because it exists.
Inputs: Candidate representations, evidence, governance decisions and acceptance authority.
Outputs: Institutionally reliable meaning available for memory, contracting and runtime use.
Canonical statement: Representation becomes institutional only when the institution accepts responsibility for relying upon it.
Institutional Memory
Category: Continuity capability
Canonical definition: The governed continuity of accepted institutional understanding, evidence, decisions, relationships, states and outcomes preserved across time, organisational change and technology evolution.
Purpose: To prevent the institution from repeatedly losing, fragmenting or silently rewriting what it knows.
What it is not: Not storage technology, a knowledge base, document repository, graph database, vector store or data lake in isolation.
Inputs: Accepted institutional representations, provenance, decisions, outcomes and revisions.
Outputs: Durable and reconstructable institutional understanding.
Canonical statement: The defining characteristic of institutional memory is continuity, not storage.
Semantic Governance Operating Model
Category: Institutional governance capability
Canonical definition: The operating model through which enterprise meaning is proposed, challenged, accepted, owned, versioned, evolved and retired.
Purpose: To establish institutional responsibility for meaning.
What it is not: Not glossary administration, data stewardship alone or a governance committee without operational authority.
Outputs: Accepted semantic architecture and governed semantic change.
Canonical statement: Meaning becomes institutional through governed acceptance.
Semantic Contract Surface
Category: Governed interoperability boundary
Canonical definition: The governed boundary through which accepted semantic architecture is exposed, bound and preserved across organisational, technical, analytical, operational and AI consumption contexts.
Purpose: To ensure accepted meaning travels without uncontrolled reinterpretation.
What it is not: Not an API alone, schema registry, ontology endpoint or data contract in isolation.
Inputs: Accepted semantic architecture, declared consumer purpose, permitted use, quality and governance obligations.
Outputs: Governed consumer binding.
Canonical statement: The Semantic Contract Surface preserves accepted meaning as it crosses boundaries.
Enterprise Memory Product
Category: Governed consumable representation
Canonical definition: A bounded, governed and reusable projection of institutional memory exposed for a declared operational, analytical, decision, workflow or AI purpose.
Purpose: To make institutional memory consumable without exposing the entire representational substrate.
What it is not: Not every table, file, topic, API, graph projection or document collection.
Inputs: Accepted institutional representation, semantic contract, provenance, purpose, ownership and permitted use.
Outputs: Purpose-bound and governed memory consumption.
Canonical statement: A memory product is a governed projection of institutional understanding.
Runtime Context Assembly
Category: Runtime intelligence capability
Canonical definition: The dynamic construction of purpose-specific context from institutional memory, current operational evidence, authority, state, constraints and execution requirements.
Purpose: To provide the right governed context for a specific actor, decision, workflow, model, agent or action.
What it is not: Not retrieval alone, prompt construction, static context storage or GraphRAG in isolation.
Inputs: Institutional memory, current evidence, purpose, actor, authority, policy and operational state.
Outputs: A bounded, traceable and purpose-specific runtime context.
Canonical statement: Context is assembled for purpose; it is not merely retrieved from storage.
Enterprise Control Plane
Category: Distributed continuity and runtime governance architecture
Canonical definition: The distributed architecture that preserves identity, meaning, authority, provenance, permitted use, lineage, conformance, accountability and reconstructability as accepted architecture becomes implementation, runtime operation, decision, action, consequence and revision.
Purpose: To preserve institutional continuity across the transition from accepted understanding to operational action.
What it is not: Not a single software platform, orchestration engine, policy tool, AI gateway or centralised command system.
Inputs: Accepted architecture, institutional memory, authority, controls, implementation state and runtime evidence.
Outputs: Conformant operational implementation and governed runtime continuity.
Canonical statement: The Enterprise Control Plane preserves institutional continuity as architecture becomes action.
Enterprise Coordination Fabric
Category: Operational coordination environment
Canonical definition: The governed environment through which people, systems, workflows, services and intelligent agents exchange context, coordinate work and participate in shared enterprise outcomes.
Purpose: To enable distributed action without losing semantic, authority or accountability coherence.
What it is not: Not an integration platform, service bus, event mesh or multi-agent framework in isolation.
Inputs: Governed context, actor identity, authority, work state, coordination intent and policy.
Outputs: Coordinated enterprise activity.
Canonical statement: Enterprise coordination is the governed movement of context, intent and work across accountable actors.
Agent Control Plane
Category: Applied runtime control architecture
Canonical definition: The applied control architecture that governs enterprise agents through identity, registry, ownership, authority, permitted use, runtime enforcement, observability, evaluation, monitoring, security, cost control, lifecycle management and offboarding.
Purpose: To make AI-enabled runtime actors governable participants in enterprise operation.
What it is not: Not the whole Enterprise Control Plane, not an agent framework and not an AI observability dashboard.
Canonical parent: Enterprise Control Plane
Canonical statement: No agent without identity, owner, authority, scope, monitoring and offboarding.
Governed Coordination
Category: Runtime operating state
Canonical definition: The condition in which people, systems, workflows and agents coordinate using accepted meaning, valid authority, traceable context and enforceable operating constraints.
Purpose: To distinguish legitimate enterprise coordination from technically possible interaction.
Outputs: Proposed decisions, workflows and actions suitable for admissibility evaluation.
Canonical statement: Coordination becomes governed when meaning, authority, purpose and accountability remain intact across actors.
Execution Admissibility
Category: Consequence-boundary determination
Canonical definition: The institutional determination that sufficient authority, state, context, constraints, risk resolution and evidence exist for a proposed consequence-bearing action to execute at T=0.
Purpose: To prevent technically executable actions from binding consequence without proven institutional integrity.
What it is not: Not decision approval, retrospective audit, model confidence or human review in isolation.
Canonical parent: Execution Admissibility Architecture
Canonical statement: No consequence-bearing state transition without proven integrity at the commit boundary.
Operational Consequence
Category: Bound institutional outcome
Canonical definition: The durable institutional outcome produced when an authorised action changes rights, obligations, state, access, ownership, records, infrastructure, funds or other consequence-bearing conditions.
Purpose: To identify the point at which action becomes institutionally real.
What it is not: Not an intention, recommendation, prediction or simulated output.
Canonical statement: Consequence is the durable institutional state created when execution binds.
Outcome Evidence
Category: Learning input
Canonical definition: The observable results, effects, state changes, exceptions and consequences produced by action and returned as evidence for evaluation.
Purpose: To determine whether institutional understanding, context, coordination and action produced the intended result.
What it is not: Not automatically a lesson, causal proof or authorised representation revision.
Outputs: Evidence available for validation, challenge and learning.
Canonical statement: Outcomes generate evidence; governance determines what the institution learns from it.
Behaviour Validation
Category: Representation assurance capability
Canonical definition: The continuous comparison of accepted institutional representations against observed data, events, relationships, behaviours and outcomes to determine whether those representations remain valid.
Purpose: To identify representational drift, contradiction, incompleteness and semantic decay.
What it is not: Not model evaluation, data quality checking or anomaly detection alone.
Outputs: Validation evidence, representation challenges and proposed revisions.
Canonical statement: Institutional understanding must remain testable against operational behaviour.
Representation Evolution
Category: Governed learning process
Canonical definition: The controlled revision, extension, supersession or retirement of institutional representations in response to new evidence while preserving provenance, continuity and prior interpretive states.
Purpose: To allow the institution to learn without silently rewriting its history.
What it is not: Not automatic model retraining, uncontrolled ontology change or direct rewriting of institutional truth by outcome data.
Inputs: Outcome evidence, behavioural validation, semantic challenge and authorised governance decisions.
Outputs: Revised and reaccepted institutional representation.
Canonical statement: Learning changes representation through governed revision, not automatic overwrite.