Audit Chen turns millions of journal entries into the few that matter — traceable, privacy-preserving and built for statutory audits under UGB and ISA.
Private study project, not a commercial offering. Access only with an approved account.
Audit Chen computes the engagement – the journal locally, Austrian and verifiable.
Journal analyses run in the browser, without upload. Documents, portal uploads and audit steps are stored on the server (EU) only if you use those features. Only when you click an AI feature do pseudonymised figures leave the device.
Austrian chart of accounts pre-set, VAT reconciliation, § 270 UGB declaration, ISA 220 (Revised) and ISQM 1 in the working papers.
A demo journal with ten planted errors is included. One click shows that all are found – and none on clean data.
Model per engagement: the same ten journal analyses plus the document request take 7.9–15.6 hours manually in Excel, around 5 minutes with Audit Chen. Reviewing the hits is needed either way and not included. Assumes a journal of around 50,000 lines.
Recently added. Every feature has an automated test that fails without it.
The document request becomes a link for the client (14 days, revocable). Uploads are matched against the journal in the browser: amount, date, supplier, requested voucher.
Preparer, reviewer, partner. Sign-off prepared → reviewed → approved, never by the same person, with an immutable trail.
Uncorrected misstatements assessed against materiality, with a suggested type of opinion (ISA 705).
Open-item lists for receivables and payables reconciled to the general ledger at year end, per personal account.
Mandatory steps per client as a progress bar, status editable in place.
Energy, Scope 1/2 and headcount from the report checked against energy, fuel and payroll expense. Every assumption visible, audit programme based on ISSA 5000.
Agent hits become a draft of risks and audit procedures. Invented audit areas, sources or numbers are rejected; the presumed ISA 240 risks are always added.
Invoice number, date, net, VAT and gross read locally from the PDF, mandatory Austrian VAT details and arithmetic checked, matched to the journal.
Only benchmarks with a reference, justification required, prior-year comparison on method change.
Audit Chen is a large data-analysis tool: 41 JET criteria, ten journal agents, sampling, matching, network analysis. So that it doesn't overwhelm you, the app ships with its own Academy. It explains not just what each button does, but why an auditor thinks this way at this point.
No customer success manager chasing meetings. No follow-up calls, no upsells. The Academy is yours: at 3 a.m. or on a Sunday on the couch, at your own pace. You figure it out yourself.
Why we do this: junior auditors spend too much time on spreadsheets, data cleansing and searching. When the machine takes over the routine, there is time for what makes the profession exciting: asking questions, seeing connections, forming a judgement.
The lessons are currently in German (the app's working language).
Request access, Academy includedExample run with demo data.
Model calculation for ten standard analyses plus the document request at around 50,000 journal lines, disclosed in the calculator above.
In the built-in self-test, each of the 41 criteria detects its planted case, and no entry of the clean test journal reaches the highest risk class. Plus Benford analysis (1st and 2nd digit, last digits) and risk scoring under ISA 240.
Posting cycles within 30 days are found automatically and shown in the FinCrime graph. The AI assessment works on pseudonymised data.
Import templates recognise the accounting system and map the columns, including debit/credit indicators. No manual mapping needed.
JET tests only the financial year plus the post-closing window. The trial balance and lead sheets pull in the prior year automatically for comparison.
Modules follow the order of the audit: planning (ISA 320), risk assessment (ISA 315), JET, substantive procedures and completion.
Every change is saved locally at once. If the database is open in another tab, Audit Chen reconnects by itself.
Detects salami slicing via cent clustering, split payments below the approval limit, repeated tiny amounts and gradual drift against the prior year (CUSUM). The journal map shows every entry as a dot in 2D. Everything runs locally.
After import, ten agents run across the journal together. Each first checks whether the data is sufficient for its task. The result is a ready-to-send document request for the client.
Animation with fictional data. Measured with a synthetic journal and deliberately planted cases per agent; runtime depends on the device.
Every decision in Audit Chen aims at an audit opinion that holds up — technically and professionally.
One connected process from data import to finding — no copying back and forth between tools.
Every figure can be traced back to its source — crucial when the oversight body asks.
Anomalies surface before they get lost under time pressure — priority instead of chance.
Samples, rationales and assessments follow the same standards — engagement after engagement.
Recurring routine work steps back so professional judgement stays in front.
No open self-service — every account is deliberately approved and can be revoked just as deliberately.
Data-analytic procedures for statutory audits under UGB and ISA — with account logic based on the Austrian standard chart of accounts. (Method pages in German.)
Risk-based analysis of all entries under ISA 240 — weekend, year-end and override entries, VAT criteria.
Digit distribution of amounts against Benford's law — indications of manipulated or invented amounts.
Overall and performance materiality from a suitable benchmark, documented traceably.
Reproducible samples under ISA 530 — monetary unit, random and risk-based selection.
Individual entries look harmless. Connected, they reveal cycles, shell companies and shared logins.
The graph shows people, companies, documents and accounts and how they connect. Step by step the pattern emerges, ending with the finding and a solution. All data here is fictional test data.
Person Company Document Account critical
Risk scoring is now computed once and reused for the table, filters, sampling and the map – with identical results.
Measured in the browser with 50,000 synthetic journal lines, same risk distribution before and after the optimisation. Rounded values, device-dependent.
When looking for money cycles, every posting has to be checked for other postings with a matching amount. It used to scan the whole pile every time. Now the amounts are kept sorted: a few targeted jumps lead straight to the handful of matching candidates.
Illustration with fictional amounts. Measured with 200,000 synthetic postings and deliberately planted cycles; values rounded, device-dependent.
| Spreadsheet | Classic audit software | Audit Chen | |
|---|---|---|---|
| Money cycles across several accounts | manual | usually via script | automatic |
| Coincidental equal amounts filtered out | – | rarely | yes |
| Hits shown as a graph | – | partly | yes, interactive |
| Data leaves the computer | no | depends on product | no |
| Installation | Office | desktop licence | browser |
General classification of typical tool categories, not a test of individual products.
The web edition runs straight in the browser. For large companies there is the Enterprise edition as a native app — with full computing power on the auditor's machine.
In the browser, no installation. Ideal for SMEs and typical annual journals.
Native app for groups, group audits and large ERP exports.
How the analysis core works in detail stays our secret. What is certain: raw data never leaves the protection layer unredacted — and the auditor always has the final say.
No small print: what is stored where, and what never leaves your device. Details in the privacy policy (German).
Access is granted on a curated basis — sign up and we will approve your account.
Request an account