Do you use any real company data?

No. The engine generates entirely from accounting rules: IFRS and GAAP standards, ERP schema architecture, and double-entry logic. There is no seed dataset at any stage, which is what makes it safe for regulated environments with no source-data exposure.

What makes the data forensic-grade?

Every debit matches a credit, the trial balance ties to the general ledger journal layer, and every key between the tables a package ships resolves, measured and printed in the package. The engine carries 37 labelled fraud schemes, with innocent look-alikes beside nearly every one: three conditions are closed sets by design (segregation of duties, the ghost vendor and phantom inventory), where the join or the reconciliation is the analytic. Each package ships a ground-truth answer key recorded at plant time, so detectors and models are scored exactly.

Which ERP platforms are covered?

SAP ECC 6.0 (BKPF/BSEG architecture), SAP S/4HANA (the ACDOCA universal journal) and Oracle Cloud (its AP, AR, GL and other module tables), plus, on the Team plan and above, a migration pair and a two-ERP intercompany group that runs as a multi-year arc.

Can I reproduce the same dataset later?

Yes, on the same engine version: the same seed and configuration regenerate a byte-identical dataset, and a package generated on an account ships a certificate naming its engine and listing the SHA-256 of each of its files. A newer engine builds a different dataset from the same seed. The watermarked public sample leaves the certificate out, because the watermark changes its PDF and workbook bytes.

Do the bank statements tie to the payments?

Yes. The statements are built from the operational cash: every payment, receipt, payroll run, remittance and tax repayment the bank posts in the year is a statement line carrying a reference to its document. The other lines are the transfers between the company's own accounts, the bank's charges and interest, and the per-bank reconciliation is proven to the cent at generation time, with the items still in transit listed.

Can I use the platform from code?

Yes, on every plan. Create an API key, install the Python client with one pip command from the evaluate page, and generate, poll and download from one terminal command or a few lines of Python, or call the REST API directly.

Is synthetic cfo an AI CFO or bookkeeping agent?

No. The name is sometimes confused with AI CFO services that automate a company's own bookkeeping or forecasting. synthetic cfo is a data engine: it generates forensic-grade synthetic ERP data for SAP and Oracle from accounting rules, with labelled fraud and a ground-truth answer key, for training and evaluating AI and audit tooling. It never touches your books.

Cold Start Engine

Checking your session…

Forensic-grade ERP data, generated from accounting rules.

synthetic cfo generates SAP and Oracle ledgers from accounting rules, with fraud planted, every planted record written down at the moment it is planted, and innocent look-alikes beside nearly every scheme, so fraud detection can be trained and measured without a single real record.

For teams that build, buy or audit fraud detection and audit analytics on SAP and Oracle.

Start free Try the demo Request access

The Developer plan is free: all three platforms, the core four cycles, a scale of up to 1,000 base transactions per cycle, and 5 datasets a month.

Already have an account? Sign in

That run produced a real package. Inspect one, before you sign up →

ERP platforms

Business cycles

Fraud schemes

100%

Reproducible per engine version

No real data, ever

Everything is generated from double-entry accounting rules, from scratch. There is no source data, so there is nothing to leak and no privacy, GDPR or re-identification risk.

An exact answer key

Every planted fraud is labelled by record, control and vector. You can score a detector's precision and recall against ground truth - the one thing real data can never give you.

Reproducible to the byte

The same seed and the same configuration rebuild the identical dataset on the same engine version, so any result can be checked by regenerating it; a newer engine builds a different dataset, and every certificate names its engine.

Real ERP architecture

SAP ECC, SAP S/4HANA and Oracle Cloud table structures, seven modules, 37 labelled fraud schemes, with double-entry holding on every line and the referential integrity measured in every package.

Start free See a real sample dataset

Every dataset ships as a four-layer package

Executive brief (whitepaper) Audit evidence + the answer key A reconciled SQLite database Source workbooks mirroring the ERP tables

What people say

Built by a chartered accountant who audited multinationals at PwC and EY and worked in SAP S/4HANA and Oracle in industry. That practitioner knowledge is encoded as deterministic generation rules - no language model ever touches the numbers.

Start free Try the demo Request access

← Overview

Sign in

Email Password

If your welcome email says so, your new account gets What's new by email, one email that gathers the releases we choose to send. You can switch it on or off any time under Notifications on your account page.

Need access? Request it here

See a real sample dataset · Try the demo

Privacy·Terms·Cookies

What should the engine build?

Describe the financial environment you need, then refine it in conversation. Built from accounting rules, never from real data.

Deterministic engine

Show me around

How is fraud embedded?

How do I score a model?

Pharma SAP environment, heavy fraud, 25,000 rows

Oracle dataset for a bank with lapping and kiting

S/4HANA migration test data, light fraud

What will I receive?

Is this reproducible?

Why synthetic data?

Deterministic engine. Your words map to a configuration you confirm. The financial data never touches a language model.

Advanced configuration

Set every parameter explicitly.

01 - ERP Platform

SAP ECC

ECC 6.0. BKPF, BSEG, EKKO, VBAK.

SAP ECC 6.0

SAP S/4HANA

Universal Journal. ACDOCA.

SAP S/4HANA

Oracle Cloud

Module tables. AP, AR, CE, GL, HCM, INV, FA.

Oracle Cloud

Migration Pair

One seed, both SAP schemas: identical subledgers, each leg its own journal layer.

ECC + S/4HANA

Intercompany Group

A holding company: SAP subsidiary + Oracle subsidiary, with elimination fraud across both.

SAP + Oracle

02 - Modules

P2P

Procure-to-Pay

O2C

Order-to-Cash

Cash Mgmt

R2R / GL

General Ledger

HR / Payroll

Insider threat

MM / Inventory

Costing fraud

FA / Fixed Assets

PP&E schemes

03 - Transaction Volume

POC

scale 500

Standard

scale 5,000

Professional

scale 25,000

Enterprise

scale 100,000

04 - Industry

Tech, Media & Telecom

SaaS, IT hardware, services.

Pharma & Life Sciences

CROs, lab equipment.

Manufacturing

Raw materials, equipment.

Financial Services

Trading, infrastructure.

Retail & Consumer

Merchandise, seasonal.

05 - Fraud Intensity

None

Pristine baseline.

Low

Subtle anomalies.

~1%

Medium

Standard training.

~3%

High

Stress-test density.

~8%

06 - Optional

Fiscal Year

Custom scale (base transactions per cycle)

Seed (optional)

Fiscal years (fraud arc)

Operational noise (entropy)

Jurisdiction

Blind track (independent fraud)

Evaluation mode (withhold markers)

Foreign currency exposure

Company size

Your jobs

No jobs yet.

Dataset insights

Live analytics computed straight from your generated databases.

Choose a dataset

Pick a completed dataset above.

Benchmark: how good is your fraud-detection AI?

Score a fraud detector, objectively

Think of it as marking an exam. We plant fraud in known places and write the answer key down as we plant it, so unlike real data, where nobody knows the true answer, we can grade any fraud-detection tool on exactly what it caught, what it missed, and where it raised a false alarm.

Dataset

See it in action

No tool of your own? Watch a fraud detector get graded on this dataset, in one click. Each is a different quality of detector, so you can see the whole range:

Simulated detectors, shown so you can see how the grading works. See a perfect score → (what submitting the answer key itself looks like).

Have your own fraud-detection tool? Grade it

Run your tool (a script, an AI model, audit software) against this dataset, then bring back the record IDs it flagged, and we find them in whatever it exported.

📁

Drop your results file here, or click to browse

CSV or TXT, exactly as your tool exported it.

Optional: only count certain fraud types:

Your account

Your identity, plan and preferences in one place.

Profile

Signs in with

Member since

Display name

Plan and usage

Plan

Generations

Includes

Generation defaults

Your usual setup. Every new chat and the Advanced panel start from these. Anything you say explicitly overrides them, and anything above your plan is fitted down with a note, never silently.

Platform

Jurisdiction

Industry

Fraud intensity

Modules

Scale (base transactions per cycle)

Appearance

System follows this device's setting.

Chats on your other devices

Keep my chats in my account so they appear on every device I sign in from. When off, chats stay only in this browser.

Notifications

Push to this device: get a system notification on your phone or computer when something happens, even with the tab closed.

What to be told about

The bell in the header keeps your newest 200 notices of the kinds you have not switched off; an announcement we take back leaves it. On iPhone, install the app first (Share, then Add to Home Screen); push reaches installed apps.

API access

Use your account from Python or any HTTP client: generate datasets, poll jobs and download packages with an API key instead of the browser. A key carries your plan and monthly allowance. It cannot manage your account, read your chats or create more keys - that stays here, behind your sign-in.

Loading…

Send it as Authorization: Bearer scfo_..., or install the Python client (the pip command on the evaluate page) and pass it to the client (also served as a single file, no dependencies). Quickstart on the evaluate page.

API playground

The code writes itself. Pick a world below and watch the exact command, Python and API call update live - then Run it now executes the same request right here on your signed-in account. No terminal, nothing installed, no key needed.

evaluation mode

Run it now counts as one generation and the dataset lands on your Datasets list with its download buttons, exactly as if the code had run on your machine. The shown code is the identical request for a pipeline, with an API key standing in for your sign-in.

Sign out and deletion

How to use synthetic cfo

From first sentence to scored model in four steps.

Quick start

Describe the environment in the Generate tab, in your own words. Platform, industry, fraud level, size - any order, any phrasing. Try: pharma SAP environment, heavy fraud, 25,000 rows

Refine in conversation: make it smaller switch to Oracle only P2P more fraud, then say generate

Download the four-layer package from the job card or History: whitepaper, audit evidence with ground-truth labels, SQL database, source workbooks. Open Insights on any completed job for live charts.

Score your model in the Benchmark tab: paste the record IDs it flagged, get precision, recall, F1 and a grade against the exact answer key.

Special builds

Migration pair: migration pair for a bank builds SAP ECC and S/4HANA from one seed: the operational subledgers are identical row for row in the two legs, and the record-to-report journal layer is generated separately for each, with its own chart, trial balance and labels (Team plan and above).

Multi-year fraud arc: retail company over 3 years, heavy fraud builds consecutive years where the schemes escalate, plus a consolidated database (Team plan and above).

Clean baseline: completely clean pharma data, no fraud produces a pristine zero-fraud control environment.

Reproducibility: every job records a seed; the same seed and configuration regenerate a byte-identical dataset on the same engine version, and a newer engine builds a different one (Reproduce says which before you press it). Set a fixed seed under Advanced.

Frequently asked questions

Live Twin

A synthetic company's ERP, posting documents in real time. Purchase orders, invoices, payments and journals stream past as the year advances - and some are fraud, hidden in the flow like they would be in a real system. The answer key stays sealed until each scheme's reveal point, so a detector is scored on catching it live, never on peeking. Freeze the world into a downloadable audit package anytime; it matches exactly what you watched.

1Set up a company and press play. It runs the same deterministic engine as the batch product.

2Watch the desk. Live figures move as documents post; fraud rides along unflagged.

3Schemes surface as the clock passes each reveal point. Seal the world into an audit package when done.

Set up the world

ERP platform Cycles Buy Sell Ledger Country Fraud level Pace

Document desk

Waiting for the first document...

Schemes surfaced released by world-time

Nothing surfaced yet. Fraud is hidden in the stream until its reveal point.

Sealing runs the engine that built this world to completion and hands you the standard four-layer package. Same seed on the same engine version, so the answer key matches exactly what streamed; a world kept across an engine update is rebuilt on the new engine.

What you receive

Every dataset ships as a four-layer package

Want to see a finished one? Open a real sample package.

Dataset insights

Computed live from your generated database

Your jobs

Running now and recently finished. Stays here after a refresh.

Chats

Your saved conversations

Notifications

What happened while you were away

We use a strictly-necessary cookie to keep you signed in, and store your chats on your device. No ad or tracking cookies. See our Cookie Notice and Privacy Policy.