📥
AI-Assisted Intake
Documents and case inputs are captured and structured. Field provenance is preserved alongside extracted values, so reviewers can trace each value back to its source.
Evidence: Operational Smoke (P0-5)
📐
Decision Checks
Decisions are evaluated against configured rules. The check is deterministic, repeatable, and produces an output that is recorded in full alongside the inputs that drove it.
Evidence: Operational Smoke (P0-5)
🗂️
Policy Trace
Each decision carries the policy version it was evaluated against and the specific rule path that produced the outcome. Policy logic is visible as part of the evidence, not buried in code.
Evidence: Operational Smoke (P0-5)
👥
Human Review
A reviewer sees the decision output, the supporting evidence, and the policy trace in one view. They approve or return the case based on what is presented, not on a separate spreadsheet.
Evidence: Operational Smoke (P0-5)
🛡️
Controlled Override
A governed exception path. Override requires a reason code from a closed vocabulary, a written rationale, a supporting evidence reference, a risk acknowledgement, and Separation of Duties (the submitter cannot approve the override on the same case). Every override is sealed into the decision record and is visible during audit replay.
Evidence: Operational Smoke (P0-5)
🔒
Frozen Artifact
On approval, the case is sealed into a decision artifact: evidence, policy result, decision output, review status, and governance metadata, recorded immutably.
Evidence: Operational Smoke (P0-5)
🛂
Execution Gate
Execution is authorised against the approved artifact. Direct action is not the trigger. Approval, attestation, token-based handoff, and role separation are all part of the gate.
Evidence: Live Atomic Consume (P0-6)
↺
Replay
Decisions can be replayed later from the stored artifact only. No live recomputation. No reliance on memory or current state. The frozen evidence is the source of truth.
Evidence: Operational Smoke (P0-5)
📊
Trust Dashboard
A business-readable view across coverage, exceptions, approval lineage, and replay evidence. Designed for managers, compliance, risk, and audit, not engineers.
Evidence: Observation Dashboard (P0-I6a)
🏢
Tenant-Scoped Governance
Configuration, policy, evidence, and replay are scoped per tenant. A single deployment can support multiple decision domains without crossing the controls between them.
Evidence: Operational Smoke (P0-5) — one tenant in pilot
⚖️
Role-Based Lifecycle
Distinct roles for submitter, reviewer, approver, and executor. The lifecycle enforces the separation; permissions are not assumed.
Evidence: Operational Smoke (P0-5)
📦
Evidence Package
Every governed decision can be exported as an evidence package: source documents, intake form, decision result, policy trace, reviewer identity, approval status, and execution record.
Evidence: Assurance Index