AI Boundary Model
ARQEN uses AI as an advisory assistant inside an evidence-first quality operating loop. It is not a black-box decision maker and does not replace responsible quality personnel.
In the platform, AI drafting is meant to reduce blank-page work across inspection plans, NCR/CAPA reviews, supplier and audit follow-up, ARQdocs packets, reporting, and management review. The output is draft text for a person to edit, apply, reject, or turn into a local Review Action. It is never autonomous approval or release.
Deterministic quality logic comes before advisory model output.
AI can draft local review text, packet sections, follow-up rationale, and summaries from source-basis context, but those drafts remain editable and human-reviewed.
AI-generated text is not compliance truth, approval, release, certification, or closure authority.
Human quality personnel retain disposition, approval, release, customer, and regulatory responsibility.
Advisory output should carry source-basis context or be marked as unavailable.
If source context, evidence references, or model routing are unavailable, ARQEN should show a blocked/unavailable draft state instead of inventing a successful output.
Production model governance, durable model-use audit, and provider-secret management require future controls.