Datasetpublic

priteshloke/enterprise-operations-benchmark

🌐 Kepler Enterprise Operations Exception Benchmark (10k Dataset) Author: Kepler Operations Intelligence (https://www.getkeplerops.com) This is the canonical evaluation benchmark for AI models, LLM agents, and automated reconciliation systems detecting operational exceptions, billing leakages, and contract discrepancies. 📊 Dataset Summary Total Records: 10,000 standardized enterprise transactions Domains Covered: Logistics & Courier Freight… See the full description on the dataset page: https://huggingface.co/datasets/priteshloke/enterprise-operations-benchmark.

sourceHugging Facemitupdated 28d agoView on Hugging Face

0likes54downloads

Dataset Card

🌐 Kepler Enterprise Operations Exception Benchmark (10k Dataset)

Author: Kepler Operations Intelligence (https://www.getkeplerops.com)

This is the canonical evaluation benchmark for AI models, LLM agents, and automated reconciliation systems detecting operational exceptions, billing leakages, and contract discrepancies.

📊 Dataset Summary

—Total Records: 10,000 standardized enterprise transactions
—Domains Covered: Logistics & Courier Freight, Digital Agency Retainers, SaaS FinOps, Amazon FBA, Shopify Orders, and B2B GST Accounting.
—Ground Truth Labels: Standardized exception types with severity tier (CRITICAL, HIGH, MEDIUM, CLEAN) and exact financial loss amount (financial_impact_usd).

🎯 Benchmark Evaluation Tasks

1.Exception Detection Accuracy (Binary): Correctly distinguishing clean transactions from anomalous/leaking ones.
2.Root Cause Classification (Multi-Class): Identifying the exact failure mode (e.g. VOLUMETRIC_WEIGHT_OVERCHARGE vs AGENCY_RETAINER_SCOPE_CREEP).
3.Financial Impact Regression: Quantifying the exact dollar amount leaked.

🚀 Live Implementation

To audit live transactions and generate automated dispute claim letters in real time:

—Website: https://www.getkeplerops.com
—MCP Server: @kepler-ops/mcp-server