synthetic cfo

0 reviews

synthetic cfo is a cold start engine for forensic-grade synthetic ERP data. It generates complete company worlds for SAP ECC, SAP S/4HANA and Oracle Cloud entirely from accounting rules: double entry, an accounting framework, statutory tax and a real ERP table architecture. No real company or personal data is used at any stage, so there is nothing to anonymise and nothing to clear with a privacy team. Every dataset is a four-layer package: an executive brief with the whitepaper and articulated financial statements, an audit evidence layer with the forensic report and the ground-truth answer key, a SQLite database with full referential integrity, and source workbooks mirroring the table architecture. Seven business cycles run on all three platforms: purchase to pay, order to cash, cash and bank, record to report, payroll, inventory and fixed assets. Ten jurisdictions set the currency, framework and tax treatment. More than 25 fraud schemes are planted per world, from ghost vendors and duplicate invoices to lapping, kiting, channel stuffing and missing-trader VAT. Each label is recorded at the moment the scheme is planted, never derived afterwards, and innocent look-alikes are planted beside nearly every scheme so a detector or a model can be measurably wrong. The same seed and configuration regenerate a package byte for byte, and a certificate records every file hash. The platform publishes the scores of real detectors against its own data, including the one that fails, and offers a free sample you can download without signing up. Access is through a web app, an API, a Python client, and an MCP server for agent frameworks. A hosted read-only ERP environment lets a practitioner walk a purchase order to the cleared bank line the way they would in the real system. Built for AI labs training and evaluating finance agents, audit and forensic teams testing detection tooling, internal audit and controls monitoring, ERP migration and testing programmes, and universities teaching forensic accounting.