Aegis Live audit ledger
New assessment

Institutional directive

AI Governance Command Center

Portfolio evaluation, risk constellation and regulatory audit preparedness for banking AI systems. Every decision logged with a cryptographic evidence hash.

Run forensic AI case View control framework

Monitored AI systems

6

100% active ingestion

Portfolio trust score

82 / 100

Target > 80

Requires action

2 models

Regulatory review pending

Audit readiness

78%

RBI & DPDP verified

01 · AI system dependency & risk constellation

Line-of-sight across every production model, its lineage and current trust posture.

High89% trust

Loan Approval Prediction Engine

Gradient Boosted Trees (LightGBM)

Production

Critical87% trust

Credit Underwriting AI v4.2

XGBoost Ensemble

Production

Critical62% trust

Fraud Detection & AML Engine

Graph Neural Network

Production

Medium91% trust

Renewal & Upsell Recommender

LightGBM

Production

High78% trust

KYC Video & Document Scorer

ResNet-50 CNN

Production

Medium84% trust

Collections Propensity Model

Survival Analysis

Production

02 · Governance & compliance drift timeline

Past 90 days

  1. Jul 15

    Re-certified +5%

    Credit Underwriting AI

    Consent lineage verified under DPDP Act 2023. Trust score moved 82% → 87%.

  2. Jul 02

    New version +8%

    Renewal Recommendation AI

    Deployed v2.1.0 with 98% SHAP explainability coverage across reason codes.

  3. Jun 20

    Validation expired −14%

    Fraud Detection AI

    Annual independent model audit lapsed. Score adjusted to 62%. CRO action required.

03 · Business impact vs AI risk quadrants

Critical risk tier

2 systems

Autonomous underwriting and fraud detection with direct lending impact.

Moderate risk & onboarding

2 systems

Video KYC scorer and collections propensity engine under human oversight controls.

Low risk / high trust

1 system

High explainability coverage with fully verified audit trails.

Onboarding queue

1 system

Loan approval engine awaiting independent challenger-model validation.

04 · Actionable insights & clause violations

View evidence vault
CriticalFraud Detection Engine · 2 hours ago

Model validation certificate expired

The annual independent review for the GNN engine lapsed. Audit readiness dropped to 58%.

Commission independent validation audit immediately.

WarningFraud Detection Engine · 5 hours ago

Explainability coverage below threshold

Graph embeddings lack reason-code decomposition for 31.8% of flagged transactions.

Deploy an integrated-gradients explainer to reach >90% coverage.

VerifiedCredit Underwriting AI · 1 day ago

Bias assessment cleared

Disparate-impact ratios within tolerance across all protected segments for Q3.

No action. Evidence sealed in the vault with hash 0x9f4c…21ab.