Inspector Configuration

MLOps / Finance IT Admin

Available Inspectors & Algorithms

Global Confidence Threshold

Settings below this threshold will be flagged for human review. Settings above will be Auto-Approved.

50%

85%

Total Detections

—

Total $ Caught

—

False Positive Rate

—

Best Inspector

—

Exact Match

Flags invoices where Company #, Vendor #, Invoice #, Date, and Amount are ALL identical.

Hits

—

$ Caught

—

FP Rate

—

Fuzzy Match

Detects similar invoice numbers (e.g., "INV-1001" vs "INV-100I") using Levenshtein distance.

Hits

—

$ Caught

—

FP Rate

—

Similarity Threshold90%

Benford's Law

Statistical analysis of leading digits to detect artificial or manipulated numerical data.

Hits

—

$ Caught

—

FP Rate

—

Weekend Posting

Flags invoices posted on Saturdays or Sundays, which is often unusual for B2B transactions.

Hits

—

$ Caught

—

FP Rate

—

Round Amount

Flags invoices with perfectly round numbers (e.g., $5,000.00), often a sign of estimates or fraud.

Hits

—

$ Caught

—

FP Rate

—

Velocity Check

Monitors for rapid-fire invoices from the same vendor within a short time window.

Hits

—

$ Caught

—

FP Rate

—

Clustering & LLM

Uses AI to cluster similar invoices and LLMs to detect semantic duplicates.

Hits

—

$ Caught

—

FP Rate

—

Semantic Vector Distance0.85

ML Anomaly Detection

Machine Learning model trained on historical data to find outliers.

Hits

—

$ Caught

—

FP Rate

—

Inspector Audit Log

Timestamp	User	Action	Inspector