# Admin Configuration Guide

*Audience: the privacy, security, AI governance, and GRC teams who set up and run the Policy Center.*

This guide covers obligations, turning policies on and tuning them, building your own policies, and reading coverage on a single entity. For the concepts behind it, start with the [Overview and Implementation Guide](/docs/policy-center/overview-and-implementation-guide/).

## Access

The Policy Center is under **Enforce → Policy Center** in the main navigation, with two tabs: **Dashboard** and **Policies**. It follows the **Data Mapping Foundation** permission scope (the same scope as Issues and Inventory). See [Configuring Scope Based Permissions](/docs/other-settings/configuring-scope-based-permissions/).

## Obligations

An obligation is a regulation, standard, or framework - GDPR, CCPA, ISO/IEC 27001, SOC 2, PCI DSS, HIPAA, the EU AI Act, ISO/IEC 42001, NIST frameworks, and many more. Each one carries the controls (clauses) Relyance's policies are mapped to.

### Enable the obligations you're held to

1. Go to **Policy Center → Policies** and click **Manage Obligations**.
2. Find each obligation by searching, or narrow the list by **Status**, **Type** (**Regulations** or **Frameworks**), **Jurisdiction**, or whether it has open issues. Switch between **List** and **Tiles**, and use **Group By** to organize the list.
3. Switch each obligation that applies to your organization **Enabled**.

Each obligation shows its jurisdiction, how many controls it has, and how many of its mapped policies are active (for example, **4 of 9 mapped policies active**).

### Review an obligation's controls

Click an obligation to open its drawer. It shows the obligation's description, identifier, and source, the **Business Entities** it's set on, and a grid of its **Controls**. For each control you can see its **Mapped Policies** - click the count to see which policies test it - and switch the control on or off.

Switching off a control you don't need stops it being enforced for every policy mapped to it.

:::warning
Disabling an obligation, or a control under it, stops enforcement for those controls. When a change would take coverage away from policies, Relyance shows **Policy Impact Detected** with the affected policies before anything is saved. Choose **Save changes anyway** to proceed. Either change can be reversed at any time.
:::

## Policies

The **Policies** tab lists every policy available to your workspace.

| Column | What it shows |
|---|---|
| **Policy** | The policy's name |
| **Default Severity** | **Low**, **Medium**, **High**, or **Critical** - the severity its issues are raised with. Editable in the grid |
| **Obligations** | The obligations whose controls the policy tests |
| **Open Issues** | How many open issues the policy has raised |
| **Status** | Whether the policy is enabled. Toggle it in the grid |

Narrow the list with the filter pills (category, policy type, obligation, severity, and status) or search, and switch between **List** and **Tiles**. The toolbar also has **Create Policy** and **Manage Obligations**.

Policies are grouped into six categories:

| Category | What its policies watch |
|---|---|
| **Data Posture** | Sensitive data exposure, encryption, masking, and secrets |
| **Identity Posture** | Access to data and systems by people and service accounts |
| **AI Security Posture** | AI models, AI agents, and MCP servers, and the data they can reach |
| **Third Party Risk** | Vendors and the data shared with them |
| **Data Residency and Sovereignty** | Where data is stored and where it flows |
| **Privacy** | Privacy-program requirements such as consent, rights requests, and records of processing |

### Turning policies on and off

Switch a policy's **Status** in the grid or in its drawer. Most **Relyance reference** policies start switched off, so you decide what to enforce. Once a policy is on, Relyance evaluates it against your environment and raises issues for anything it matches.

:::warning
Disabling a policy stops enforcement for its controls and closes its existing open issues. Relyance asks you to confirm with **Yes, deactivate policy**. You can turn it back on at any time.
:::

### Reading a policy

Click a policy to open its drawer. From top to bottom it shows:

- **What it watches** - the kind of record the policy checks and the conditions that make something an issue, written out as a sentence.
- **Controls** - the controls it tests, grouped by framework. Use **Disable control only for this policy** to stop a single policy enforcing a control without disabling the control everywhere.
- **How to remediate** - the policy's remediation steps, each with **What**, **Why**, and **How**.
- **Risks** - the [Risk Register](/docs/risk-register/overview-and-implementation-guide/) risks this policy monitors, and library risks whose controls map to it, which you can **Add to register** from here.
- **Issues** - the issues the policy has raised.

A details rail alongside shows the policy's **Category**, **Severity**, **Status**, **Origin** (**Relyance reference** or **Custom**), **Created**, **Last run**, and **Coverage**.

### Editing and cloning

You can edit your own custom policies directly with **Edit Policy**. A **Relyance reference** policy can be switched on or off, rescored, and have individual controls disabled, but its conditions can't be changed: click **Clone to edit** to start a new custom policy from a copy of it.

### Re-running evaluation

Relyance evaluates enabled policies automatically, and a new or edited policy starts evaluating as soon as it's saved. To force a full refresh, open the **Policies** tab's menu and choose **Run policy evaluation**. This evaluates every policy and recomputes all issues for your workspace, and can take a few minutes.

## Creating a policy

Click **Create Policy** on the **Policies** tab. The builder has three steps: **Define**, **Configure**, and **Review**. Anything you've already passed stays open, so you can step back and change it.

### 1. Define what to watch

1. **Choose what to watch** - the kind of record the policy checks: **Data assets**, **Identities**, **Access grants**, **AI models**, **AI agents**, **MCP servers**, **Services**, **Repositories**, **Vendors**, **Domains**, **Data requests**, or **Assessments**.
2. Add the conditions that make one of those records an issue. Each condition picks a field and a test - for example **is any of**, **is none of**, **is recorded**, **is not recorded**, **is older than**, **is within the last**, or **is inside (by hierarchy)**. A record is raised when it matches all of the conditions.
3. Or click **Define with AI** and describe the policy in one sentence. Relyance writes the conditions for you and names anything in your sentence it couldn't turn into a condition.

As you build, **If it ran today…** previews how many records the policy would raise, out of how many it watches. A policy that matches nothing yet is still worth saving - it will raise an issue the first time something matches.

### 2. Configure the policy

- **Policy Name**, **Category**, and **Severity**. Relyance suggests a name and category from your conditions; use **Suggest a name** to ask again.
- **Controls** - the framework controls this policy tests. Relyance recommends controls based on what the policy watches, and the counter shows how many you've selected against how many it recommends.
- **Title** - the wording each issue is raised with, for example "Unmasked data in `{violation.entity.assetType}` `{violation.entity.name}`". Template variables fill in the details of each record; **Template variables** lists the ones you can use. Use **Suggest a title** for a draft.
- **Remediation steps** - a one-sentence summary of what to do, then steps with **What to do** and **Why it matters**. Use **Suggest steps** for drafts based on similar policies. A step can be set to open a **Jira** ticket, a **GitHub** issue, or a **Slack** message from the issue.
- **Enabled** - whether the policy starts running as soon as it's saved.

### 3. Review and create

The review page shows the whole policy in one read, with the fields Relyance wrote for you marked **Written for you** so you can check them. Click **Review and create** to save. An enabled policy starts evaluating as soon as it's saved.

If your new policy asks exactly the same thing as one that already exists, Relyance won't create a duplicate and offers to open the existing policy instead.

:::info
Where enabled for your workspace, the **Asset Explorer**, **Identities**, and AI inventory pages also offer **Create a policy from this filter**, which opens the builder with that page's filters already in place.
:::

## Coverage on a single entity

The drawers for data assets, identities, vendors, repositories, AI models, AI agents, and MCP servers have a **Policies & Coverage** tab. It shows every policy that watches that entity and the controls they monitor on it. Switch between **By policy** and **By control**.

| Standing | Meaning |
|---|---|
| **Passing** | The policies that check this control found no issue on this entity |
| **Failing** | At least one policy found an issue - use **View risk** to open it |
| **Not decided** | The policy applies, but there isn't enough data to judge it yet - for example, no data flows seen, or a value it asks about is missing |

A control is failing if any policy that monitors it fails on the entity.

## Configuration checklist

- [ ] Enable every obligation your organization is held to under **Manage Obligations**.
- [ ] Review each enabled obligation's controls and switch off any that don't apply.
- [ ] Turn on the reference policies that check your enabled obligations.
- [ ] Set each policy's **Default Severity** to match your priorities.
- [ ] Check the dashboard's **Coverage gaps** card until every enabled obligation has its mapped policies on.
- [ ] Build custom policies for organization-specific rules, with controls and remediation steps.
- [ ] Link policies to the risks they monitor in the [Risk Register](/docs/risk-register/admin-configuration-guide/#monitoring-a-risk).
