Intelligence Value Objects¶
Why This Exists¶
This document defines immutable and value-based concepts used by the Intelligence Domain.
Value objects protect meaning for confidence, approval state, reasoning type, memory scope, and AI provenance.
Owner¶
The owner is the Chief Product Officer and Enterprise Architect.
Business Value¶
Precise value objects improve trust, auditability, and consistent AI behavior.
Value Object Catalogue¶
| Value object | Purpose | Validation rules |
|---|---|---|
| IntelligenceSessionId | Identifies a reasoning session. | Required, unique, immutable. |
| AgentId | Identifies registered AI agent. | Required for agent-executed work. |
| DigitalProfessionalId | Identifies Digital Professional role. | Must map to approved Digital Professional. |
| OrganizationContextRef | References Organization context used. | Must include OrganizationId and context version where available. |
| ConfidenceScore | Represents model or reasoning confidence. | Numeric range, method, threshold, and context required. |
| ConfidenceBand | Human-readable confidence classification. | Controlled values such as low, medium, high. |
| ReasoningObjective | Describes reasoning purpose. | Must be business-readable. |
| ReasoningMode | Type of reasoning used. | Controlled values such as classification, extraction, recommendation, planning, review. |
| EvidenceReference | Points to evidence used. | Must include source domain and source ID where possible. |
| ExplainabilitySummary | Concise rationale for output. | Must include evidence, assumptions, and limitations. |
| ApprovalRequirement | Defines approval need. | Controlled values such as not required, recommended, required. |
| GuardrailResult | Policy result. | Controlled values such as allowed, blocked, requires approval, requires review. |
| MemoryScope | Defines memory visibility. | Organization, domain, session, or global policy scope. |
| AuditCorrelationId | Correlates intelligence actions. | Required for multi-step reasoning or execution. |
Confidence Scoring Rule¶
Confidence scores are decision support signals, not guarantees. Any score used in customer-facing recommendations must be explainable and tied to evidence quality.
Organization Context Rule¶
OrganizationContextRef is mandatory when Intelligence reasons about customer-bound procurement work.