Assess and Baseline

Conduct comprehensive AI discovery and risk assessment to establish quantitative baseline for QAG implementation

Phase 1: Week 8 of 12

Assessment Progress: 67% Complete

Phase 1 Implementation Progress (Weeks 1-12)

AI Inventory

Discover and catalog all AI/ML assets across the enterprise

Models Discovered 247

Completion 85%

On Track

Risk Audit

Comprehensive risk assessment using QAG taxonomy

Models Assessed 163/247

Completion 66%

In Progress

Tiger Team

Cross-functional team formation and pilot selection

Team Members 7/8

Completion 88%

Nearly Complete

Assessment Summary

3.2

Current Governance Maturity (Out of 5.0)

Critical Risk Models

23 High-Risk

Action Required

Shadow AI Discovered

41 Ungoverned

Priority Focus

Compliance Gap

$2.4M Risk Exposure

Quantified

AI Model Inventory & Risk Heat Map

Model Name	Business Unit	Model Type	Risk Score	Status	Discovery Method
Credit-Risk-Scorer-v3	Financial Services	Classification	High (8.2)	Production	Code Scanning
Customer-Churn-Predictor	Marketing	Classification	Medium (5.7)	Staging	Survey Response
Fraud-Detection-Engine	Security	Anomaly Detection	High (7.9)	Production	Cloud Discovery
Product-Recommender-v2	E-commerce	Collaborative Filtering	Low (3.2)	Production	Code Scanning
Sentiment-Analyzer	Customer Service	NLP Classification	Medium (4.8)	Development	Survey Response
Price-Optimization-Model	Sales	Regression	Medium (6.1)	Production	Code Scanning
Resume-Screening-AI	Human Resources	Classification	High (8.7)	Pilot	Survey Response
Supply-Chain-Optimizer	Operations	Optimization	Low (2.9)	Production	Cloud Discovery

Risk Distribution

23

High Risk

89

Medium Risk

135

Low Risk

Discovery Methods

Code Scanning 142

Survey Response 78

Cloud Discovery 27

Baseline Metrics Establishment

Current State Assessment

Metric	Current State	Target	Gap
Model Documentation Coverage	23%	100%	Large
Bias Testing Coverage	12%	100%	Critical
Performance Monitoring	34%	95%	Large
Explainability Implementation	8%	80%	Critical
Risk Assessment Completion	66%	100%	Medium
Governance Automation	5%	85%	Critical
Audit Trail Completeness	41%	100%	Large
Stakeholder Training	18%	90%	Critical

Pilot Project Selection & Criteria

Selection Criteria

High business visibility

Manageable risk scope

Clear success metrics

Executive sponsor

3-6 month timeline

Cross-functional impact

Top Pilot Candidates

Credit Risk Scoring Model

High-visibility, regulated, clear ROI metrics

Selected

Customer Recommendation Engine

Medium visibility, manageable complexity

Alternative

HR Resume Screening

High bias risk, good learning opportunity

Backup

Technology Stack Assessment: Build vs. Buy Analysis

Component	Build Score	Buy Score	Recommendation	Rationale
Governance Dashboard	6.2	8.4	Buy	Mature vendor solutions available, faster time-to-value
Risk Scoring Engine	8.1	5.7	Build	Highly organization-specific, competitive advantage
Model Registry	4.3	8.9	Buy	Standard functionality, proven solutions (MLflow)
Policy-as-Code Engine	7.8	6.2	Build	Custom rules needed, integration complexity
Bias Detection Tools	3.4	9.1	Buy	Excellent open source options (FairLearn, AIF360)
Explainability Framework	4.1	8.6	Buy	Mature libraries (SHAP, LIME), standardized
Monitoring & Alerting	5.8	7.9	Buy	Infrastructure complexity, vendor expertise

Recommended Hybrid Approach

Buy: Core Platform

Governance dashboard, monitoring

Build: Custom Logic

Risk scoring, policy rules

Leverage: Open Source

MLflow, FairLearn, SHAP

Key Vendor Candidates

DataRobot Governance

IBM OpenPages

SAS Model Risk Management

Evidently AI