Intelligence
ARQEN's intelligence layer separates source truth, deterministic interpretation, advisory assistance, human review, and evidence traceability. Quality signals can be traced back to their records instead of being treated as unexplained AI output.
Source truth
Signals begin with source records: NCRs, CAPAs, inspections, audits, supplier events, documents, reviews, and traceability context.
Deterministic interpretation
Rule-based readiness, lifecycle gates, missing evidence, owner/due status, and blocked states are evaluated before advisory AI is involved.
Advisory assistance
AI can summarize, explain, and suggest next review questions, but it does not approve, release, certify, or close regulated work.
Human review
Quality personnel keep authority over disposition, approval, release, and customer or regulatory commitments.
Evidence traceability
Every useful signal should be traceable back to source-basis context or honestly marked as unavailable.
| Truth | Coverage | Primary metric | Delta vs prior | Process condition |
|---|
This public readout separates observed source context from delayed or reference-only context. Rows marked as reference-only are useful for product walkthroughs, but they are not production evidence, not live operational telemetry, and not approval authority. The strip above summarizes source coverage, provenance mix, reliability, operational impact, and compact risk-attention counts without hiding where source context is unavailable.
Technical implementation details belong in docs. For how this layer fits the platform, see Intelligence layer docs and the product overview.