Inventory Discrepancy Audit

Use observation tools to isolate problematic SKUs:

Tools: get_observed_inventory_snapshot
Parameters: {quality_threshold: 0.7, include_anomalies: true}

Decision Point: If multiple products show discrepancies, prioritize by value (ABC analysis).

Step 2: Validate Sensor Health

Check devices contributing to inventory observations:

Tools: list_sensor_devices, get_device_health_summary
Focus: anomaly_count, last_calibration, signal_quality

Known Patterns:

• anomaly_count > 5 in 24 hours → Likely hardware issue
• signal_quality < 0.8 → Check calibration or positioning
• No recent calibration → Schedule maintenance

Step 3: Trace Inventory Movements

Audit recent transactions for the affected product:

Tools: list_inventory_moves, get_inventory_move_audit_trace
Parameters: {product_id: "<SKU>", start_date: "<T-7days>"}

Red Flags:

• Moves without corresponding receipts/shipments
• Adjustments without approval codes
• Unusual transfer patterns (e.g., high-value items to low-security zones)

Step 4: Cross-Reference Procurement Data

Verify expected quantities against purchase orders:

Tools: list_purchase_orders, get_purchase_order
Filter: status="received", product_id="<SKU>"

Check: received_quantity vs. ordered_quantity — partial deliveries can cause count mismatches.

Step 5: Calculate Discrepancy Severity

Apply statistical reasoning to quantify the issue:

Shrinkage Rate Formula:

shrinkage_rate = (expected_qty - observed_qty) / expected_qty * 100

Confidence Interval (95%):

import numpy as np
from scipy.stats import norm

expected = 1000
observed = 950
std_dev = 15  # historical variance

z_score = (expected - observed) / std_dev
p_value = norm.sf(abs(z_score)) * 2

if p_value < 0.05:
    conclusion = "Statistically significant discrepancy — investigate further"