GenAI DQS Agent

Universal Payment Data Quality Scoring

System Ready

Upload Payment Data

Upload individual files or entire folders for batch analysis.

⚡ Simulate AI Failure (Circuit Breaker)

Composite Quality Score

0.0 / 100

Analyzing...

GenAI Executive Summary

Our agent has processed your dataset. Please review the detailed dimension breakdown below. The system has applied the "Rules > ML" logical containment to ensure safety.

Why This is NOT a Black Box

Full Transparency & Explainability

✓ Formula-Based Scoring

All 7 dimensions use explicit mathematical formulas, not learned weights.

✓ Rules > ML

Machine Learning is advisory only. Deterministic rules always take authority.

✓ Auditable Logs

Every decision is logged with timestamp, inputs, outputs, and rationale.

📐 Scoring Formulas (Fully Transparent)

Completeness:100 × (1 - nulls/total)

Accuracy:100 × (1 - issues/rows)

Validity:100 × (1 - invalid/checks)

Uniqueness:100 × (1 - dupes/rows)

🔬 ML Methods (Explainable, Not Deep Learning)

→

Z-Score Outlier Detection: z = (value - mean) / std. Flagged if |z| > 2.5σ

→

Category Analysis: Compares value to category-specific baseline statistics

→

Temporal Patterns: Detects burst activity exceeding 2σ from daily mean

⚙️ Decision Logic (Finite State Machine)

Decision	Condition	Human Override
SAFE_TO_USE	DQS ≥ 75%, Anomalies ≤ 5, No violations	Allowed
REVIEW_REQUIRED	DQS ≥ 50%, OR Anomalies > 5	Required
ESCALATE	DQS < 30%, OR Critical violations	Mandatory

Responsibility Boundary & Audit Trace

REAL-TIME LOGS