AI Governance Readiness Assessment

Assess your organization's AI governance maturity with results saved to SerenDB.

What This Skill Provides

• Structured intake questionnaire for organization context
• Maturity scoring across 8 governance domains (1-5 scale)
• Framework alignment (NIST AI RMF, EU AI Act, ISO 42001, OCC SR 11-7)
• Gap analysis with prioritized recommendations
• Board-ready assessment reports
• SerenDB persistence for historical tracking and follow-up

Runtime Files

• scripts/serendb_schema.sql - Database schema for assessment persistence

When to Use

Activate this skill when the user asks about:

• "AI governance assessment"
• "assess our AI readiness"
• "AI compliance check"
• "governance maturity evaluation"
• "NIST AI RMF assessment"
• "EU AI Act readiness"
• "view past assessments"
• "track governance progress"

SerenDB Setup (First Run)

Before running assessments, initialize the database schema via MCP:

1. Resolve or create project:

mcp__seren-mcp__list_projects
mcp__seren-mcp__create_project(name: "ai-governance")

2. Create database:

mcp__seren-mcp__create_database(project_id: "<project_id>", name: "governance_assessments")

3. Apply schema:

mcp__seren-mcp__run_sql(
  project_id: "<project_id>",
  database: "governance_assessments",
  query: "<contents of scripts/serendb_schema.sql>"
)

Workflow

Phase 1: Organization Context

Ask these intake questions:

1. Industry: Financial Services, Healthcare, Energy/Utilities, Government/Defense, Technology, Manufacturing, Other
2. Size: <100, 100-1000, 1000-10000, 10000+ employees
3. AI Maturity: Exploring (no production AI), Early (1-5 use cases), Scaling (5-20), Mature (20+)
4. Regulatory Environment: GDPR/EU AI Act, US Federal, Financial (OCC/SEC), Healthcare (HIPAA/FDA), Industry-specific
5. Current State: AI Center of Excellence, Ethics policy, Model inventory, Risk assessment process, Human oversight, Bias testing

Phase 2: Framework Selection

Based on responses, determine applicable frameworks:

| Trigger | Framework | |---------|-----------| | EU operations | EU AI Act | | US Federal/Government | NIST AI RMF | | Financial services | OCC SR 11-7 | | Healthcare | FDA AI/ML guidance | | Any AI deployment | NIST AI RMF (baseline) |

Phase 3: Domain Assessment

Evaluate maturity (1-5) across 8 domains:

1. Strategy & Leadership - Board oversight, AI governance body, executive sponsorship
2. Risk Management - AI risk inventory, assessment process, escalation procedures
3. Data Governance - Training data documentation, quality standards, consent management
4. Model Lifecycle - Versioning, approval workflows, drift detection, retirement
5. Transparency - Explainability, user disclosure, regulatory documentation
6. Fairness & Bias - Pre-deployment testing, protected attributes, remediation
7. Security & Privacy - AI-specific security, adversarial testing, privacy techniques
8. Human Oversight - High-risk review, override capability, operator training

Scoring Guide:

• 1: No governance, ad-hoc
• 2: Informal, reactive
• 3: Documented, emerging
• 4: Formal, proactive
• 5: Continuous, automated

Phase 4: Gap Analysis

For each domain:

1. Calculate current score
2. Determine target based on risk profile
3. Identify gap (target - current)
4. Prioritize: CRITICAL (gap 3+, high urgency), HIGH, MEDIUM, LOW

Phase 5: Save to SerenDB

After completing assessment, persist results via MCP:

1. Insert main assessment record:

INSERT INTO governance.assessments (
  organization_name, industry, employee_count, ai_maturity,
  overall_score, maturity_level, frameworks_applicable, assessor
) VALUES (
  'First Regional Bank', 'financial_services', '1000-10000', 'early',
  2.4, 'developing', ARRAY['occ_sr_11_7', 'nist_ai_rmf', 'fair_lending'], 'AI Governance Skill'
) RETURNING assessment_id;

2. Insert domain scores:

INSERT INTO governance.domain_scores (
  assessment_id, domain_name, current_score, target_score, gap, priority, notes
) VALUES
  ('<assessment_id>', 'strategy_leadership', 2.0, 4.0, 2.0, 'high', 'No board oversight'),
  ('<assessment_id>', 'risk_management', 2.0, 4.0, 2.0, 'critical', 'No model inventory'),
  -- ... all 8 domains
;

3. Insert identified gaps:

INSERT INTO governance.gaps (
  assessment_id, gap_title, domain_name, priority,
  current_state, target_state, risk_description,
  recommendation, effort, timeline_days
) VALUES (
  '<assessment_id>',
  'No AI/ML Model Inventory',
  'risk_management',
  'critical',
  '3 AI systems deployed with no centralized tracking',
  'Comprehensive inventory with risk classifications',
  'OCC examination deficiency, potential enforcement action',
  'Catalog all AI/ML systems using standardized template',
  'medium',
  30
);

4. Insert risk register entries:

INSERT INTO governance.risk_register (
  assessment_id, risk_id, category, description,
  likelihood, impact, risk_score, current_controls, recommended_actions
) VALUES (
  '<assessment_id>',
  'AI-001',
  'fairness',
  'Credit scoring model may have disparate impact',
  'high',
  'high',
  9,
  'None',
  'Immediate bias testing, ongoing monitoring'
);

5. Insert roadmap items:

INSERT INTO governance.roadmap_items (
  assessment_id, phase, phase_name, item_description, target_days
) VALUES
  ('<assessment_id>', 1, 'Foundation', 'Complete AI/ML model inventory', 30),
  ('<assessment_id>', 1, 'Foundation', 'Conduct bias testing on credit model', 60),
  ('<assessment_id>', 2, 'Build', 'Deploy model registry with approval workflows', 180);

Phase 6: Generate Report

Output structured report (see examples/sample-assessment-regional-bank.md).

Querying Past Assessments

View all assessments:

SELECT * FROM governance.v_assessment_summary ORDER BY assessment_date DESC;

View specific assessment details:

SELECT a.*, d.domain_name, d.current_score, d.target_score, d.priority
FROM governance.assessments a
JOIN governance.domain_scores d ON d.assessment_id = a.assessment_id
WHERE a.organization_name = 'First Regional Bank'
ORDER BY d.priority;

Track gap remediation progress:

SELECT * FROM governance.v_gap_progress
WHERE organization_name = 'First Regional Bank';

Compare assessments over time:

SELECT organization_name, assessment_date, overall_score, maturity_level,
       critical_gaps, high_gaps
FROM governance.v_assessment_summary
WHERE organization_name = 'First Regional Bank'
ORDER BY assessment_date;

Update gap status:

UPDATE governance.gaps
SET status = 'resolved', resolved_at = NOW(), resolution_notes = 'Model inventory completed'
WHERE assessment_id = '<id>' AND gap_title = 'No AI/ML Model Inventory';

Example

User: "Assess our AI governance. We're a 2,000-employee regional bank with 3 AI systems."

Agent workflow:

1. Ask clarifying questions (AI maturity, regulations, current governance)
2. Determine frameworks (OCC SR 11-7, NIST AI RMF, Fair Lending)
3. Score each domain based on responses
4. Generate gap analysis with priorities
5. Save results to SerenDB via MCP SQL calls
6. Output board-ready report with roadmap
7. Provide assessment_id for future reference

Follow-up: "Show me our governance progress since last assessment"

• Query governance.v_assessment_summary for historical scores
• Query governance.v_gap_progress for remediation tracking

See examples/sample-assessment-regional-bank.md for full output.

Target Users

• Chief Compliance Officers
• AI/ML Governance Program Managers
• Risk Leaders in Regulated Industries
• Enterprise Architects evaluating AI adoption

About Benjamin Black Consulting

This skill is provided by Benjamin Black Consulting, specializing in AI strategy and governance advisory for regulated industries.

• AI Governance Frameworks
• Data Strategy & Architecture
• Compliance Readiness Assessments
• Executive AI Training