Intelligence Entities¶
Why This Exists¶
This document defines Intelligence Domain entities with identity.
Entities track intelligence work over time, including agents, sessions, memories, approvals, and audits.
Owner¶
The owner is the Chief Product Officer and Enterprise Architect.
The Intelligence Domain owns the entity definitions listed here.
Business Value¶
Entity clarity supports governed AI behavior, auditability, explainability, and integration with Organization and execution workflows.
Entity Catalogue¶
| Entity | Purpose | Key attributes | Business rules | Relationships | MVP status |
|---|---|---|---|---|---|
| Intelligence Director | Coordinates intelligence routing and orchestration. | DirectorId, status, routing policy version. | Must use registered agents only. | Uses Agent Registry and Reasoning Sessions. | MVP |
| AI Agent | Runtime AI identity or agent. | AgentId, name, role, status, allowed capabilities. | Must be registered and governed. | May represent Digital Professional runtime. | MVP |
| Digital Professional Runtime | Operational runtime for a Digital Professional. | RuntimeId, professional name, agent ID, Practice scope. | Must map to approved Digital Professional. | Uses Agent Registry and guardrails. | MVP |
| Reasoning Session | Bounded reasoning activity. | SessionId, OrganizationId, objective, context, status. | Must record inputs, output, confidence, explanation. | Produces memory, events, audit. | MVP |
| Reasoning Step | Step inside a reasoning session. | StepId, session ID, type, input, output. | Must preserve sequence and provenance. | Belongs to Reasoning Session. | Later |
| AI Memory Entry | Intelligence memory record. | MemoryId, OrganizationId, source, content summary, retention. | Must include provenance and scope. | References sessions and events. | MVP |
| Execution Request | Request to execute or advance workflow. | RequestId, OrganizationId, requested action, status. | Requires guardrail evaluation. | May be accepted by Execution Engine. | MVP |
| Guardrail Decision | Human approval or automation policy result. | DecisionId, policy, result, approver, reason. | Required for controlled execution. | Belongs to Execution Request or Session. | MVP |
| Explainability Record | Explanation for AI output or action. | ExplanationId, session ID, evidence, rationale, limitations. | Must reference evidence used. | Linked to Reasoning Session. | MVP |
| Confidence Assessment | Confidence scoring record. | AssessmentId, score, method, threshold, limitations. | Must not be presented as certainty. | Linked to output or recommendation. | MVP |
| Audit Log Entry | Immutable record of intelligence activity. | AuditId, actor, action, timestamp, correlation ID. | Must be append-only. | Links sessions, agents, approvals. | MVP |
| Event Subscription | Configured event interest. | SubscriptionId, event type, consumer, status. | Must be governed. | Connects to Event Mesh. | Later |
Entity Rule¶
All customer-bound intelligence entities must include OrganizationId or a traceable relationship to an Organization-bound event.