Docs
relyance.ai ↗
Browse docs

Legal and Compliance Guide

Updated November 4, 2026View as Markdown ↗

Audience: legal, privacy, security, compliance, and audit teams. No UI steps - for those, see the Admin Configuration Guide.

Why continuous control monitoring

Most frameworks expect more than a written policy: they expect you to show that your controls operate in practice and to find out when they stop. GDPR's accountability principle (Article 5(2)) and security-of-processing requirement (Article 32), ISO/IEC 27001's monitoring and measurement clause (9.1), SOC 2's monitoring criteria, PCI DSS, the HIPAA Security Rule, and the EU AI Act's post-market monitoring for high-risk AI systems all ask, in different words, for ongoing evidence rather than a point-in-time attestation.

The Policy Center provides that evidence by testing your obligations' controls against what Relyance discovers in your connected environment - data assets, identities, vendors, services, repositories, AI models, AI agents, and MCP servers - and raising an issue wherever a control doesn't hold.

What the Policy Center demonstrates

You need to show What the Policy Center gives you
You know which obligations apply An explicit, reviewable list of enabled obligations, with controls you've deliberately switched off
Controls are mapped to checks Every policy maps to the specific clauses it tests, grouped by framework
Controls operate in practice Enabled policies evaluate continuously against your environment, and each control has a standing per entity: passing, failing, or not decided
Gaps are visible The Coverage gaps card lists enabled obligations whose mapped policies aren't all on
Failures are acted on Each issue carries remediation steps and can open a ticket in your tracker; issue status shows what's open, in progress, and closed
Coverage of a specific system An entity's Policies & Coverage tab shows every policy that watches it and each control's standing

What it doesn't show

Keep these limits in mind when you present Policy Center results as evidence:

  • Scope is what Relyance can see. Policies evaluate the systems you've connected through integrations. A system that isn't connected isn't checked.
  • No open issues is not the same as compliant. It means the enabled policies found nothing to raise within their scope. Many obligations also include organizational, contractual, and procedural requirements that no technical check can test.
  • Not decided is not passing. A control marked Not decided on an entity means there wasn't enough data to judge it yet - for example, no data flows observed.
  • Your choices shape the result. Which obligations are enabled, which controls and policies are switched off, and each policy's severity are your organization's decisions. Review them periodically.

AI review of issues

Relyance AI can review an issue and record a verdict - Considerable risk or Not a considerable risk - with its key signal, an explanation, and a trace of the steps it took. Keep in mind:

  • The review only closes an issue as a false positive when its evidence was moderate or strong; weaker reviews leave the issue open.
  • The default Issues view shows considerable risks the review is highly confident about, plus unreviewed issues. Issues the review was less sure of are kept, under the Unconfirmed or low confidence filter, not discarded - include them when you need a complete picture for an audit.
  • The summary and trace are the review's own record of its reasoning. Treat them as supporting evidence that a person can check, not as a substitute for that judgment.

See Issues and AI Review.

Policies you write

Custom policies, including ones drafted with Define with AI, are reviewed by a person before they're saved: the builder marks every field Relyance wrote so it can be checked. Relyance won't save a policy that asks exactly the same thing as an existing one, which keeps the catalogue free of duplicate checks.