Skip to content

Control 2.12: Supervision and Oversight (FINRA Rule 3110)

Overview

Control ID: 2.12 Control Name: Supervision and Oversight (FINRA Rule 3110) Regulatory Reference: FINRA Rule 3110, FINRA Rule 4511, SEC 17a-4, SOX 302/404 Setup Time: 2-3 hours initial setup, ongoing supervision

Purpose

This control establishes supervisory procedures for AI agents in broker-dealer and financial services operations, directly aligned with FINRA Rule 3110's supervision requirements. FINRA Rule 3110 requires member firms to establish and maintain written supervisory procedures reasonably designed to achieve compliance with applicable laws and regulations. When AI agents are deployed for customer-facing or business-critical functions, the firm must designate qualified supervisory personnel, implement appropriate oversight mechanisms, and maintain supervision records. This control defines the supervisory structure, review procedures, and documentation requirements for AI agent oversight.


Prerequisites

Primary Owner Admin Role: Compliance Officer Supporting Roles: AI Governance Lead

Required Licenses

  • Microsoft 365 E3/E5 (for compliance features)
  • Power Platform per-user or per-app licenses
  • Copilot Studio

Required Permissions

  • Compliance Officer (supervision program ownership)
  • AI Governance Lead (day-to-day oversight)
  • Chief Compliance Officer (ultimate supervisory responsibility)

Dependencies

  • Control 2.8 (Access Control and Segregation of Duties)
  • Control 1.7 (Audit Logging)
  • Control 3.1 (Agent Inventory)

Pre-Setup Checklist

  • [ ] Supervisory personnel designated
  • [ ] Written supervisory procedures drafted
  • [ ] Review cadence defined
  • [ ] Escalation procedures established
  • [ ] Training on AI supervision completed

Governance Levels

Baseline (Level 1)

Document supervision procedures for agent oversight; define approver roles and responsibilities.

Quarterly supervision reports; documented approval workflow for agent deployments.

Regulated/High-Risk (Level 4)

Real-time supervision dashboard; mandatory oversight by Compliance Officer; supervision records retention.


Setup & Configuration

Step 1: Establish Supervisory Structure

Define the supervisory hierarchy for AI agents per FINRA 3110.

AI Agent Supervisory Hierarchy:

┌─────────────────────────────────────────────────────────────────┐
│                    CHIEF COMPLIANCE OFFICER (CCO)               │
│              Ultimate supervisory responsibility                │
│          Quarterly review of AI supervision program             │
└─────────────────────────────────┬───────────────────────────────┘
                                  │
              ┌───────────────────┴───────────────────┐
              │                                       │
┌─────────────▼─────────────┐         ┌───────────────▼───────────────┐
│   AI GOVERNANCE LEAD       │         │   DESIGNATED SUPERVISOR(S)    │
│  Day-to-day program mgmt   │         │   By business line/function   │
│  Monthly performance review│         │   Weekly operational review   │
└─────────────┬─────────────┘         └───────────────┬───────────────┘
              │                                       │
              └───────────────────┬───────────────────┘
                                  │
                    ┌─────────────▼─────────────┐
                    │      AGENT OWNERS          │
                    │   Responsible for agent    │
                    │   performance & content    │
                    └───────────────────────────┘

Step 2: Document Written Supervisory Procedures (WSP)

Create WSP section for AI Agent governance per FINRA 3110.

AI Agent WSP Template:

WRITTEN SUPERVISORY PROCEDURES
Section: AI Agent Governance

1. PURPOSE
This section establishes supervisory procedures for the use of AI agents
(including Copilot Studio agents) in firm operations, consistent
with FINRA Rule 3110.

2. SCOPE
These procedures apply to all AI agents that:

- Interact with customers
- Access customer or firm data
- Provide recommendations or advice
- Execute transactions or processes
- Generate communications

3. SUPERVISORY PERSONNEL DESIGNATION
3.1 Chief Compliance Officer: [Name] - Ultimate oversight
3.2 AI Governance Lead: [Name] - Program management
3.3 Designated Supervisors by Function:
    - Customer Service Agents: [Name]
    - Investment Agents: [Name]
    - Operations Agents: [Name]

4. PRE-DEPLOYMENT REVIEW
4.1 All enterprise-managed agents require CCO approval before production
4.2 Review must include:
    - Compliance with applicable regulations
    - Accuracy of information provided
    - Appropriate disclosures
    - Escalation paths to human personnel
    - Data protection measures

5. ONGOING SUPERVISION
5.1 Daily: Automated monitoring of error rates and escalations
5.2 Weekly: Designated supervisor review of sample interactions
5.3 Monthly: AI Governance Lead performance review
5.4 Quarterly: CCO program review

6. SAMPLE REVIEW REQUIREMENTS
6.1 Enterprise-managed agents: Review [X]% of interactions weekly
6.2 Team collaboration agents: Review [X]% of interactions monthly
6.3 Document findings in supervision log

7. ESCALATION PROCEDURES
7.1 Customer complaints regarding AI: Immediate supervisor notification
7.2 Regulatory inquiries: Escalate to CCO within 24 hours
7.3 Material errors or failures: Immediate suspension and review

8. RECORD KEEPING
8.1 Maintain all supervision records for minimum 6 years
8.2 Records include: approvals, reviews, findings, actions taken

9. ANNUAL REVIEW
9.1 WSP for AI agents reviewed annually by CCO
9.2 Updates documented and communicated to supervisory personnel

Step 3: Create Agent Approval Workflow

Build Power Automate workflow for supervisory approval.

Copilot Studio Agent Approval Flow:

  1. Navigate to make.powerautomate.com
  2. Create Automated cloud flow
  3. Trigger: When agent status changes to "Ready for Review"
  4. Flow steps:

Enterprise Managed Agents:

1. Get agent details from registry
2. Send approval to Designated Supervisor
   - Include: Agent summary, test results, risk assessment
3. If approved: Send to AI Governance Lead
4. If approved: Send to CCO (Chief Compliance Officer)
5. Require all three approvals
6. If all approved: Update status to "Approved for Production"
7. Log approval chain with timestamps

Team Collaboration Agents:

1. Get agent details from registry
2. Send approval to Team Lead
3. Send approval to AI Governance Lead
4. Require both approvals
5. Update status and log approvals

Step 4: Configure Supervision Dashboard

Build Power BI dashboard for supervisory oversight.

Supervision Dashboard Components:

Section Content Refresh
Agent Overview Total agents by governance tier, status, owner Daily
Interaction Summary Volume, resolution rate, escalations Real-time
Error Monitoring Error count, rate, trending Real-time
Sample Review Status Reviews completed vs. required Weekly
Approval Pipeline Pending approvals, aging Daily
Compliance Issues Open issues, remediation status Daily
Supervision Log Recent supervision activities Real-time

Step 5: Implement Sample Review Process

Define and automate interaction sampling for review.

Sample Selection Criteria:

Trigger Selection Logic Priority
Random X% of all interactions Standard
Keyword Contains sensitive terms High
Escalation Required human handoff High
Low CSAT Customer rating < 3 High
New Topic First occurrence of topic Medium
Error Agent error detected High

Create SharePoint List: Supervision Review Log

Column Type Purpose
Review ID Auto-number Unique identifier
Agent ID Lookup Link to agent registry
Session ID Text Specific interaction
Review Date Date When review conducted
Reviewer Person Designated supervisor
Selection Reason Choice Random, Keyword, Escalation, etc.
Interaction Summary Multi-line text Brief description
Accuracy Choice Accurate, Minor issues, Inaccurate
Compliance Choice Compliant, Concern, Violation
Customer Impact Choice Positive, Neutral, Negative
Follow-up Required Yes/No Needs further action
Follow-up Action Multi-line text Description of action
Resolution Choice Closed, Pending, Escalated

Step 6: Establish Escalation Procedures

Define clear escalation paths for supervision issues.

Escalation Matrix:

Issue Type Initial Contact Escalation 1 Escalation 2 Timeline
Customer complaint about AI Agent Owner AI Governance Lead CCO 24 hours
Incorrect information provided Designated Supervisor AI Governance Lead CCO 48 hours
Regulatory concern identified AI Governance Lead CCO Legal Immediate
Material agent failure IT Operations AI Governance Lead CCO 4 hours
Data breach/exposure Security Team CCO + CISO Executive Team Immediate
Bias/discrimination concern AI Governance Lead CCO + Legal Executive Team 24 hours

Step 7: Configure Automated Supervision Alerts

Power Automate Alert Flow:

  1. Create scheduled flow (every 4 hours)
  2. Check for:
  3. Agents without review in > 7 days
  4. Error rates exceeding threshold
  5. Pending approvals > 48 hours old
  6. Unresolved escalations
  7. Send consolidated alert to supervisory personnel

Step 8: Document Supervision Records

Ensure all supervision activities are logged for FINRA 4511 compliance.

Required Supervision Records:

  • Agent approval documentation
  • Sample review logs
  • Supervisory meeting minutes
  • Issue identification and resolution
  • Training completion records
  • Annual WSP review documentation
  • Examination response records

Retention Requirements:

  • FINRA 4511: Minimum 6 years
  • SEC 17a-4: 6 years for books and records
  • Firm policy: Consider 7+ years for AI governance

PowerShell Configuration

# ============================================================
# Control 2.12: Supervision and Oversight (FINRA Rule 3110)
# ============================================================

# This script supports supervision program configuration
# and generates compliance documentation

# -------------------------------------------------------------
# Section 1: Define Supervisory Structure
# -------------------------------------------------------------

Write-Host "Configuring AI Agent Supervisory Structure..." -ForegroundColor Cyan

$SupervisoryStructure = @{
    CCO = @{
        Role = "Chief Compliance Officer"
        Responsibilities = @(
            "Ultimate supervisory responsibility for AI agents",
            "Quarterly program review",
            "Annual WSP review and approval",
            "Regulatory examination response"
        )
        ReviewCadence = "Quarterly"
    }
    AIGovernanceLead = @{
        Role = "AI Governance Lead"
        Responsibilities = @(
            "Day-to-day program management",
            "Monthly performance reviews",
            "Approval workflow management",
            "Supervision log oversight"
        )
        ReviewCadence = "Monthly"
    }
    DesignatedSupervisor = @{
        Role = "Designated Supervisor"
        Responsibilities = @(
            "Weekly operational review",
            "Sample interaction review",
            "First-level approval for changes",
            "Issue identification and escalation"
        )
        ReviewCadence = "Weekly"
    }
    AgentOwner = @{
        Role = "Agent Owner"
        Responsibilities = @(
            "Agent performance monitoring",
            "Content accuracy maintenance",
            "Incident response",
            "Change request initiation"
        )
        ReviewCadence = "Daily"
    }
}

$SupervisoryStructure | ConvertTo-Json -Depth 3 | Out-File "Supervisory_Structure.json"
Write-Host "Supervisory structure exported" -ForegroundColor Green

# -------------------------------------------------------------
# Section 2: Generate WSP Template
# -------------------------------------------------------------

Write-Host "`nGenerating Written Supervisory Procedures template..." -ForegroundColor Cyan

$WSPTemplate = @"
===============================================================================
WRITTEN SUPERVISORY PROCEDURES - AI AGENT GOVERNANCE
Firm: [FIRM NAME]
Effective Date: [DATE]
Last Reviewed: [DATE]
Approved By: [CCO NAME]
===============================================================================

SECTION 1: PURPOSE AND SCOPE
-----------------------------
1.1 Purpose
These procedures establish the supervisory framework for AI agents deployed
within [Firm Name] operations, consistent with FINRA Rule 3110 requirements
for establishing, maintaining, and enforcing written supervisory procedures.

1.2 Scope
These procedures apply to all AI agents including:

- Copilot Studio agents
- Custom conversational AI implementations
- Automated customer communication systems
- AI-assisted advisory tools

1.3 Regulatory References
- FINRA Rule 3110: Supervision
- FINRA Rule 4511: Books and Records
- SEC Rule 17a-4: Records Preservation
- FINRA Notice 25-07: AI in Broker-Dealer Operations

SECTION 2: SUPERVISORY PERSONNEL
---------------------------------
2.1 Chief Compliance Officer
Name: [CCO NAME]
Responsibilities: Ultimate supervisory authority, quarterly review, WSP approval

2.2 AI Governance Lead
Name: [NAME]
Responsibilities: Program management, monthly reviews, approval workflow

2.3 Designated Supervisors
[Business Function 1]: [NAME]
[Business Function 2]: [NAME]
[Business Function 3]: [NAME]

SECTION 3: PRE-DEPLOYMENT SUPERVISION
-------------------------------------
3.1 Agent Classification
All agents must be classified into a governance tier:

- Personal productivity: Personal/Low risk - Team lead approval
- Team collaboration: Team/Medium risk - AI Governance Lead + Designated Supervisor
- Enterprise managed: Enterprise/High risk - Full approval chain including CCO

3.2 Pre-Deployment Review Checklist
[ ] Agent purpose and use case documented
[ ] Regulatory implications assessed
[ ] Data access scope reviewed
[ ] Testing completed (functional, security, bias)
[ ] Disclosures verified (AI usage disclosure)
[ ] Escalation paths to human personnel confirmed
[ ] Approval workflow completed per governance tier

SECTION 4: ONGOING SUPERVISION
-------------------------------
4.1 Daily Supervision
- Automated monitoring of error rates and system health
- Agent Owner review of any alerts or issues

4.2 Weekly Supervision
- Designated Supervisor sample review of interactions
- Minimum review rates:
    Enterprise managed: [X]% of interactions
    Team collaboration: [Y]% of interactions
    Personal productivity: [Z]% random sample

4.3 Monthly Supervision
- AI Governance Lead performance review
- Aggregated metrics analysis
- Issue trending and root cause review

4.4 Quarterly Supervision
- CCO program review
- Supervisory procedure effectiveness assessment
- Regulatory update integration

SECTION 5: SAMPLE REVIEW PROCEDURES
------------------------------------
5.1 Selection Methodology
Samples selected based on:

- Random selection (statistical sampling)
- Keyword/trigger-based (risk-weighted)
- Customer complaint association
- Escalation events
- Low satisfaction scores

5.2 Review Documentation
Each review must document:

- Interaction accuracy assessment
- Compliance with firm policies
- Regulatory adherence
- Customer impact evaluation
- Follow-up actions required

5.3 Review Record Retention
All review records retained minimum 6 years per FINRA 4511.

SECTION 6: ESCALATION PROCEDURES
---------------------------------
6.1 Issue Categories and Response
- Customer Complaints: Notify supervisor within 24 hours
- Compliance Concerns: Escalate to CCO within 24 hours
- Regulatory Inquiries: Immediate CCO notification
- Material Failures: Suspend agent pending review

6.2 Escalation Documentation
All escalations documented including:

- Issue description
- Immediate actions taken
- Resolution timeline
- Follow-up verification

SECTION 7: RECORD KEEPING
--------------------------
7.1 Required Records
- Agent approval documentation
- Supervision review logs
- Escalation records
- Training completion records
- WSP amendments

7.2 Retention Period
Minimum 6 years from creation or last applicable use.

7.3 Format
Records maintained in firm's designated recordkeeping systems
with appropriate access controls and backup.

SECTION 8: ANNUAL REVIEW
-------------------------
8.1 Annual WSP Review
This WSP section reviewed annually by CCO or designee.
Review includes:

- Effectiveness assessment
- Regulatory update incorporation
- Best practice updates
- Personnel updates

8.2 Documentation
Annual review documented with:

- Review date
- Changes made
- Approver signature

===============================================================================
VERSION HISTORY
---------------
| Version | Date       | Author     | Changes                              |
|---------|------------|------------|--------------------------------------|
| 1.0     | [Date]     | [Name]     | Initial version                      |
| 2.0     | [Date]     | [Name]     | [Changes]                            |

===============================================================================
CERTIFICATION

I have reviewed and approved these Written Supervisory Procedures for AI
Agent Governance.

Chief Compliance Officer: _________________________ Date: ___________

===============================================================================
"@

$WSPTemplate | Out-File "WSP_AI_Agent_Template.txt"
Write-Host "WSP template generated" -ForegroundColor Green

# -------------------------------------------------------------
# Section 3: Create Supervision Log Schema
# -------------------------------------------------------------

Write-Host "`nGenerating supervision log schema..." -ForegroundColor Cyan

$LogSchema = @"

SUPERVISION LOG - DATAVERSE/SHAREPOINT SCHEMA
==============================================

Table: fsi_supervisionlog

| Column                | Type         | Description                          |
|-----------------------|--------------|--------------------------------------|
| fsi_reviewid          | Auto-number  | Unique review identifier             |
| fsi_agentid           | Lookup       | Link to Agent Registry               |
| fsi_sessionid         | Text         | Specific interaction ID              |
| fsi_reviewdate        | Date/Time    | When review was conducted            |
| fsi_reviewer          | Lookup       | Link to User                         |
| fsi_reviewerRole      | Choice       | CCO, AI Gov Lead, Designated Sup     |
| fsi_selectionreason   | Choice       | Random, Keyword, Escalation, etc.    |
| fsi_interactionsummary| Multi-line   | Brief description of interaction     |
| fsi_accuracy          | Choice       | Accurate, Minor Issues, Inaccurate   |
| fsi_compliance        | Choice       | Compliant, Concern, Violation        |
| fsi_customerimpact    | Choice       | Positive, Neutral, Negative          |
| fsi_followuprequired  | Yes/No       | Needs further action                 |
| fsi_followupaction    | Multi-line   | Description of required action       |
| fsi_followupassigned  | Lookup       | Person responsible for follow-up     |
| fsi_followupduedate   | Date         | When follow-up must be complete      |
| fsi_resolution        | Choice       | Closed, Pending, Escalated           |
| fsi_resolutiondate    | Date/Time    | When issue was resolved              |
| fsi_resolutionnotes   | Multi-line   | Resolution details                   |

Create in Power Apps Maker Portal -> Tables -> New table

"@

Write-Host $LogSchema -ForegroundColor Yellow

# -------------------------------------------------------------
# Section 4: Generate Supervision Metrics Report
# -------------------------------------------------------------

Write-Host "`nGenerating supervision metrics report template..." -ForegroundColor Cyan

$MetricsReport = @"
===============================================================================
QUARTERLY SUPERVISION METRICS REPORT
Period: [Q# YYYY]
Report Date: $(Get-Date -Format 'yyyy-MM-dd')
Prepared By: [AI Governance Lead]
Reviewed By: [CCO]
===============================================================================

EXECUTIVE SUMMARY
-----------------
This report summarizes AI agent supervision activities for the quarter,
demonstrating compliance with FINRA Rule 3110 supervisory requirements.

SUPERVISION COVERAGE
--------------------
| Tier   | Active Agents | Reviews Completed | Review Rate | Target |
|--------|---------------|-------------------|-------------|--------|
| Enterprise managed | [X] | [Y] | [Z]% | [T]% |
| Team collaboration | [X] | [Y] | [Z]% | [T]% |
| Personal productivity | [X] | [Y] | [Z]% | [T]% |

INTERACTION SAMPLING
--------------------
Total Interactions This Quarter: [X]
Interactions Reviewed: [Y]
Sample Rate: [Z]%

Sampling Distribution:

- Random Selection: [X]%
- Keyword Triggered: [Y]%
- Escalation Events: [Z]%
- Low CSAT: [W]%

FINDINGS SUMMARY
----------------
| Category          | Count | Resolved | Open | Resolution Rate |
|-------------------|-------|----------|------|-----------------|
| Accuracy Issues   | [X]   | [Y]      | [Z]  | [%]             |
| Compliance Concerns| [X]  | [Y]      | [Z]  | [%]             |
| Customer Impact   | [X]   | [Y]      | [Z]  | [%]             |
| Escalations       | [X]   | [Y]      | [Z]  | [%]             |

APPROVAL ACTIVITY
-----------------
New Agent Approvals: [X]
Agent Modifications Approved: [Y]
Approvals Denied: [Z]
Average Approval Time: [Days]

ESCALATION SUMMARY
------------------
Total Escalations: [X]
To Designated Supervisor: [Y]
To AI Governance Lead: [Z]
To CCO: [W]

Escalation Reasons:
[List top escalation reasons]

SUPERVISION MEETINGS
--------------------
Weekly Operations Reviews Held: [X] of [Y] scheduled
Monthly Performance Reviews Held: [X] of [Y] scheduled
Quarterly CCO Reviews Held: [X] of [Y] scheduled

KEY ISSUES AND ACTIONS
----------------------
[List significant issues identified and actions taken]

RECOMMENDATIONS
---------------
[List recommendations for program improvement]

CERTIFICATION
-------------
I certify that this supervision program has been executed in accordance
with our Written Supervisory Procedures and FINRA Rule 3110 requirements.

AI Governance Lead: _________________ Date: _________
Chief Compliance Officer: _________________ Date: _________

===============================================================================
"@

$MetricsReport | Out-File "Quarterly_Supervision_Report_Template.txt"
Write-Host "Quarterly report template generated" -ForegroundColor Green

# -------------------------------------------------------------
# Section 5: FINRA 3110 Compliance Checklist
# -------------------------------------------------------------

Write-Host "`nFINRA Rule 3110 AI Supervision Compliance Checklist..." -ForegroundColor Cyan

$Checklist = @"
===============================================================================
FINRA RULE 3110 - AI AGENT SUPERVISION COMPLIANCE CHECKLIST
===============================================================================

WRITTEN SUPERVISORY PROCEDURES (WSP):

[ ] WSP includes specific procedures for AI agent supervision
[ ] WSP designates responsible supervisory personnel
[ ] WSP defines approval requirements by governance tier/risk
[ ] WSP establishes review cadence and sampling methodology
[ ] WSP specifies escalation procedures
[ ] WSP reviewed and updated at least annually
[ ] WSP approved by CCO with documented signature

SUPERVISORY PERSONNEL:

[ ] CCO designated with ultimate AI oversight responsibility
[ ] AI Governance Lead designated for program management
[ ] Designated Supervisors assigned by business function
[ ] All supervisory personnel completed AI supervision training
[ ] Supervisory structure documented and communicated

PRE-DEPLOYMENT SUPERVISION:

[ ] All enterprise-managed agents require CCO approval before production
[ ] All team collaboration agents require AI Governance Lead approval
[ ] Approval workflow documented and enforced
[ ] Pre-deployment checklist completed for each agent
[ ] Testing results reviewed before approval

ONGOING SUPERVISION:

[ ] Sample review conducted per WSP requirements
[ ] Review findings documented in supervision log
[ ] Follow-up actions tracked to resolution
[ ] Supervision meetings held per schedule
[ ] Metrics reported to supervisory personnel

ESCALATION:

[ ] Escalation procedures documented and communicated
[ ] Escalations tracked and resolved
[ ] CCO notified of material issues
[ ] Regulatory inquiries handled per procedure

RECORD KEEPING:

[ ] All supervision records maintained per FINRA 4511
[ ] Records accessible for regulatory examination
[ ] Retention policy enforced (minimum 6 years)
[ ] Records protected from unauthorized modification

ANNUAL REVIEW:

[ ] WSP reviewed annually by CCO
[ ] Changes documented and approved
[ ] Supervisory personnel updated on changes
[ ] Training refreshed as needed

===============================================================================
"@

Write-Host $Checklist -ForegroundColor Yellow

Write-Host "`nSupervision configuration complete" -ForegroundColor Green

Financial Sector Considerations

Regulatory Alignment

Regulation Supervision Requirement Control Implementation
FINRA Rule 3110 Establish and maintain supervisory procedures Written WSP for AI agents
FINRA Rule 3110(b) Designate registered principal for each activity Supervisory hierarchy defined
FINRA Rule 4511 Maintain supervisory records 6+ year retention of logs
SEC 17a-4 Preserve books and records Supervision records retained
SOX 302/404 Management certification of controls CCO quarterly review
FINRA Notice 25-07 AI-specific supervision guidance Bias testing, disclosure oversight

Zone-Specific Configuration

Zone Supervision Level Approval Chain Sample Rate
Zone 1 - Personal Light touch Team Lead 1% random
Zone 2 - Team Standard AI Gov Lead + Supervisor 5% + keyword
Zone 3 - Enterprise Intensive Full chain + CCO 10% + keyword + all escalations

FSI Supervision Considerations

Broker-Dealer Agents:

  • FINRA supervision requirements apply
  • Suitability determination oversight
  • Best execution monitoring
  • Communications supervision per Rule 3110(b)(4)

Investment Adviser Agents:

  • Fiduciary duty oversight
  • Advice accuracy supervision
  • Disclosure completeness review

Banking Agents:

  • Consumer protection oversight
  • Fair lending supervision
  • Privacy compliance monitoring

Verification & Testing

Verification Steps

  1. WSP Documented
  2. Review AI Agent WSP section
  3. Verify CCO approval signature
  4. Confirm annual review date current

  5. Supervisory Personnel Designated

  6. Review organizational structure
  7. Verify personnel trained on AI supervision
  8. Confirm coverage for all business functions

  9. Approval Workflow Enforced

  10. Test sample approval request
    • Verify approval chain per governance tier
  11. Check that unapproved agents blocked

  12. Sample Reviews Conducted

  13. Check supervision log entries
  14. Verify sample rates met
  15. Confirm findings addressed

Compliance Checklist

  • [ ] Written Supervisory Procedures include AI agents
  • [ ] Supervisory personnel designated and trained
  • [ ] Approval workflow enforced per governance tier
  • [ ] Sample reviews conducted per schedule
  • [ ] Supervision dashboard operational
  • [ ] Escalation procedures documented
  • [ ] Supervision records retained 6+ years
  • [ ] Quarterly CCO review conducted
  • [ ] Annual WSP review completed

Troubleshooting & Validation

Issue: Approval Workflow Not Enforced

Symptoms: Agents deployed without proper approval Solution:

  1. Review approval workflow configuration
  2. Ensure workflow triggers on correct status change
  3. Verify approvers have appropriate access
  4. Add validation to prevent status bypass
  5. Audit existing agents for approval gaps

Issue: Sample Review Targets Not Met

Symptoms: Review rates below WSP requirements Solution:

  1. Increase automation of sample selection
  2. Add review queue notifications
  3. Redistribute workload across supervisors
  4. Consider risk-based prioritization
  5. Escalate staffing needs if persistent

Issue: Supervision Records Missing

Symptoms: Unable to produce records for examination Solution:

  1. Audit all supervision activities
  2. Verify logging is capturing all reviews
  3. Check retention policy enforcement
  4. Reconstruct records where possible
  5. Implement controls to prevent recurrence

Issue: CCO Not Engaged in AI Supervision

Symptoms: Quarterly reviews not occurring Solution:

  1. Schedule recurring calendar appointments
  2. Prepare executive summary reports
  3. Highlight regulatory risk of non-compliance
  4. Delegate preparation to AI Governance Lead
  5. Document and escalate if persistent

Additional Resources

Control Relationship
2.8 - Access Control Supervisors need appropriate access
1.7 - Audit Logging Log supervision activities
2.11 - Bias Testing Supervision includes bias oversight
2.13 - Documentation Supervision records retention
3.3 - Compliance Reporting Supervision metrics in reports

Support & Questions

For implementation support or questions about this control, contact:

  • AI Governance Lead (governance direction)
  • Compliance Officer (regulatory requirements)
  • Technical Implementation Team (platform setup)

Updated: Dec 2025
Version: v1.0 Beta (Dec 2025)
UI Verification Status: ❌ Needs verification