Skip to main content

AI proposals

Vantage can use AI to suggest governance work — mapping a policy to a control, flagging an uncovered domain, drafting a document or a redline.

Nothing an agent produces is applied automatically. Every suggestion arrives as a proposal that a person accepts or rejects. That gate is the whole design: the AI drafts, a human decides, and the decision is recorded.

Proposal types

TypeWhat it suggests
Control mappingAttach a policy to a framework or risk control.
Coverage action"There is no governing document for domain X."
Framework scaffoldCreate a policy plus its child procedures.
Policy draftA full bilingual document draft.
Policy redlineA proposed diff against the current version.

Deciding

Each proposal is pending until someone acts:

  • Accept — applied, and recorded as your decision.
  • Reject — dismissed, with the reasoning kept.
  • Needs info — sent back for more context.
  • Withdrawn — retracted by whoever raised it.

Accepted and rejected are terminal. Re-opening a decided proposal is not supported by design: the audit trail should show what was decided and when, so a changed mind creates a new proposal rather than rewriting the old one.

The run record

Every agent run is recorded — the model used, the steps it took and what it produced. If a regulator asks how a policy came to be drafted, that trail is the answer.

AI output is a draft, not an authority

A generated policy has not been reviewed by your legal or compliance function until someone does that. Accepting a draft proposal creates a document in draft — it still has to go through review, approval and publication like any other.