> ## Documentation Index
> Fetch the complete documentation index at: https://docs.encoreos.io/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Compliance Advisor - Administrator Guide

> Module: GR Core - AI Compliance Advisor (GR-06) Audience: Compliance Officers, System Administrators Last Updated: 2025-12-14

**Module:** GR Core - AI Compliance Advisor (GR-06)\
**Audience:** Compliance Officers, System Administrators\
**Last Updated:** 2025-12-14

***

## Overview

This guide covers administration of the AI Compliance Advisor feature, including enabling AI, managing suggestions, reviewing AI outputs, and configuring settings.

***

## Enabling AI Features

### Prerequisites

* Organization administrator or compliance officer role
* Access to GR module settings

### Enable AI for Your Organization

1. Navigate to **GR → Settings**
2. Click the **AI** tab
3. Toggle **Enable AI Compliance Advisor** to ON
4. Click **Save Changes**

Once enabled, AI features will be available to all GR module users.

### Disable AI Features

To disable AI for your organization:

1. Navigate to **GR → Settings → AI**
2. Toggle **Enable AI Compliance Advisor** to OFF
3. Click **Save Changes**

> **Note:** Disabling AI does not delete existing AI-generated data. Users will retain access to previously generated suggestions.

***

## Managing AI Suggestions

### Review Workflow

As a compliance officer, you are responsible for reviewing AI-generated suggestions before they are implemented.

**Review Process:**

1. Navigate to **GR → AI → Suggestions**
2. Filter by **Status: Pending** to see unreviewed suggestions
3. Click on a suggestion to review details
4. Take action:
   * **Accept**: Marks suggestion as approved, may create related records
   * **Reject**: Dismisses suggestion with a documented reason
   * **Defer**: Saves for later review

### Bulk Actions

For multiple suggestions:

1. Select suggestions using checkboxes
2. Use bulk action buttons:
   * **Accept Selected**
   * **Reject Selected** (prompts for reason)

### Suggestion Lifecycle

```
AI Generated → Pending → Review
                           ↓
                    ┌──────┴──────┐
                    ↓             ↓
               Accepted       Rejected
                    ↓
               Implemented
```

***

## Reviewing AI Gap Analyses

### Gap Analysis Workflow

1. **Generate**: User or scheduled job triggers gap analysis
2. **Review**: Compliance officer reviews findings
3. **Validate**: Confirm or adjust gap severity/priority
4. **Action**: Create remediation tasks or dismiss false positives

### Managing Gap Analyses

1. Navigate to **GR → AI → Gap Analysis**
2. View analyses by requirement or date
3. For each gap:
   * Verify accuracy against actual compliance status
   * Adjust priority if needed
   * Create remediation action or dismiss

### Creating Remediation from Gaps

1. Open a gap analysis result
2. Click **Create Remediation** on any gap item
3. The system pre-fills:
   * Description from gap analysis
   * Suggested actions as tasks
   * Priority based on gap priority
4. Assign owner and due date
5. Save to create remediation record

***

## Audit Preparation Management

### Managing AI Audit Prep

1. Navigate to **GR → AI → Audit Prep**
2. View preparation assessments by audit
3. For each assessment:
   * Review readiness score and narrative
   * Validate checklist items
   * Assign checklist items to staff
   * Track completion

### Readiness Score Interpretation

| Score     | Status    | Action Needed                 |
| --------- | --------- | ----------------------------- |
| 90-100%   | Excellent | Minor polish, ready for audit |
| 70-89%    | Good      | Address priority items        |
| 50-69%    | Fair      | Significant work needed       |
| Below 50% | Poor      | Major remediation required    |

### Exporting Audit Prep

* Click **Export** to generate a PDF or Excel checklist
* Use for team distribution and tracking
* Include in audit preparation documentation

***

## AI Risk Assessment Management

### Managing AI Risk Assessments

1. Navigate to **GR → AI → Risk Assessment**
2. View assessments by risk
3. For each assessment:
   * Compare AI scores with current risk scores
   * Validate or adjust likelihood/impact
   * Review suggested mitigation strategy
   * Create mitigation actions if appropriate

### Applying AI Recommendations

When accepting an AI risk assessment:

1. Open the assessment details
2. Click **Apply to Risk**
3. Select which values to update:
   * Likelihood score
   * Impact score
   * Mitigation strategy
4. Add any manual adjustments
5. Save to update the risk record

***

## AI Prompt Templates

The AI uses specialized prompts for different functions. Understanding these helps interpret outputs.

### Compliance Chat

* General compliance guidance
* Regulatory interpretation
* Documentation requirements
* Best practices

### Requirement Identification

Identifies applicable requirements based on:

* Organization type
* Accreditations held
* State of operation
* Services provided

### Gap Analysis

Analyzes gaps considering:

* Evidence gaps (missing documentation)
* Check gaps (missing verifications)
* Training gaps (staff training needs)
* Policy gaps (missing/outdated policies)

### Audit Preparation

Assesses readiness based on:

* Audit type (CARF, State, CMS)
* Historical findings
* Current compliance status
* Documentation completeness

### Risk Assessment

Evaluates risks using:

* Likelihood (1-5 scale)
* Impact (1-5 scale)
* Risk matrix calculation
* Mitigation strategy recommendations

***

## AI Model Information

The AI Compliance Advisor uses Lovable AI, which provides access to advanced language models.

### Models Used

* **Default**: Google Gemini 2.5 Flash (balanced speed/quality)
* Complex analysis may use higher-capability models

### Model Capabilities

* Regulatory knowledge (CARF, Joint Commission, CMS, HIPAA, SAMHSA)
* State licensing requirements
* Compliance best practices
* Gap identification
* Risk assessment methodology

### Model Limitations

* Knowledge may not include latest regulatory updates
* Cannot access external databases or real-time data
* Provides guidance, not legal advice
* May not know state-specific variations for all states

***

## Best Practices

### Getting Accurate AI Results

1. **Provide Context**: When using AI chat, include relevant details:
   * Organization type (recovery housing, behavioral health)
   * Accreditations (CARF, Joint Commission)
   * State of operation
   * Specific services offered

2. **Be Specific**: Ask targeted questions rather than broad ones:
   * ✅ "What are CARF documentation requirements for medication management?"
   * ❌ "Tell me about CARF"

3. **Verify Outputs**: Always validate AI suggestions against:
   * Current regulatory standards
   * State-specific requirements
   * Your organization's specific context

4. **Use Confidence Scores**: Pay attention to confidence levels:
   * High confidence (80%+): Likely accurate
   * Lower confidence: Requires more validation

### Reviewing AI Suggestions Efficiently

1. **Prioritize by Type**: Review in this order:
   * Critical priority items first
   * Requirement suggestions (new compliance needs)
   * Gap suggestions (current deficiencies)
   * Remediation suggestions (action items)

2. **Batch Similar Items**: Group related suggestions for efficient review

3. **Document Rejections**: When rejecting, provide clear reasons to:
   * Inform future AI improvements
   * Maintain audit trail
   * Help other reviewers understand decisions

***

## Data Privacy

### What Data is Sent to AI

* Anonymized organizational context
* Regulatory framework references
* Compliance questions and queries
* **Never**: PHI, PII, or identifying information

### Data Protection

* All AI interactions are logged for audit purposes
* No patient/resident identifiable information is included
* AI suggestions are stored in your organization's database
* Data remains within your organization's tenant

### PHI Protection

The system automatically:

* Uses anonymized references (e.g., "Resident R-001")
* Strips identifying information from prompts
* Blocks queries that appear to contain PHI

***

## Rate Limits & Usage

### Understanding Rate Limits

AI features have usage limits to ensure fair access:

| Error                 | Meaning             | Resolution              |
| --------------------- | ------------------- | ----------------------- |
| 429 Too Many Requests | Rate limit exceeded | Wait a moment and retry |
| 402 Payment Required  | Credits depleted    | Contact administrator   |

### Managing Usage

* Monitor usage through the AI dashboard
* Prioritize AI use for high-value activities
* Use batch operations when possible
* Cache results rather than regenerating

***

## Troubleshooting

### Common Issues

| Issue                    | Cause               | Solution                          |
| ------------------------ | ------------------- | --------------------------------- |
| AI features not visible  | Not enabled         | Enable in GR Settings → AI        |
| "Rate limited" error     | Too many requests   | Wait and retry                    |
| "Credits depleted" error | Usage limit reached | Contact workspace admin           |
| Inaccurate suggestions   | Missing context     | Provide more organization context |
| Slow responses           | High demand         | Wait for completion, don't retry  |

### AI Not Working

1. Verify AI is enabled: **GR → Settings → AI**
2. Check for error messages in the UI
3. Try a simple test query in AI Chat
4. Check network connectivity
5. Contact system administrator if issues persist

### Reporting AI Issues

To report problems with AI outputs:

1. Note the specific suggestion/analysis ID
2. Document what was incorrect
3. Provide the expected correct answer if known
4. Submit feedback through the suggestion interface

***

## Module Settings Reference

### AI Settings Tab

| Setting                      | Description                       | Default |
| ---------------------------- | --------------------------------- | ------- |
| Enable AI Compliance Advisor | Master toggle for all AI features | OFF     |

### Related Settings

Other GR settings that affect AI behavior:

* Organization type and accreditations (contextual)
* State of operation (regulatory context)
* Services offered (requirement filtering)

***

## Security Considerations

### Access Control

* All org members can view AI suggestions
* All org members can generate AI content
* Only compliance officers can accept/reject suggestions
* Only compliance officers can delete AI data

### Audit Trail

All AI interactions are logged including:

* User who triggered generation
* Timestamp
* Input context
* Generated output
* Accept/reject decisions with reasons

### Data Retention

AI-generated data follows your organization's data retention policies. To remove AI data:

1. Navigate to the AI dashboard
2. Select items to delete
3. Confirm deletion (compliance officer only)

***

## Related Documentation

* [AI User Guide](./ai-compliance-user-guide.md)
* [AI Prompts Reference](./ai-prompts-reference.md)
* [Compliance Admin Guide](./compliance-admin-guide.md)
* [Audit Admin Guide](./audit-admin-guide.md)
* [Risk Admin Guide](./risk-admin-guide.md)