Industries/Financial services/Audit
AI consulting for audit firms
An audit file is built from documents that the client sends in every format. Staff request invoices, contracts, bank confirmations and minutes and match each one to a sample item by hand. They rebuild the same workpapers from last year's file. Journal testing starts with a full ledger export in a spreadsheet, and review notes are cleared one at a time before the file is assembled.
Talk to our team →- Use cases
- Audit evidence matching, Document search
- Runs on
- Your servers, on premises
- Connects to
- Caseware, Caseware IDEA, DataSnipper, TeamMate+
- Rules cited
- 05, each with its source
- Decides
- A person, every time
Architecture
We build beside the software you already run
We work with mid-tier and smaller registered company auditors, SMSF auditors and internal audit teams. The model runs on your own servers or in an Australian cloud region, and no client file goes to a public AI service. The system prepares and indexes the evidence and flags items for review. The auditor forms the opinion and signs the audit report.
- SourcesCaseware / Caseware IDEA / DataSnipper / TeamMate+ / CCH iFirm
- Your serversAudit evidence matching
- ReviewYour team
- OutputEvidence requests / Evidence matching / Sampling support / Journal testing / Workpapers and review notes
The hours in audit go to work that repeats
- 01
Evidence requests
Request lists built for each client from last year's file, then invoices, contracts, bank confirmations and minutes chased by email before fieldwork can start.
- 02
Sample testing
Each sample item matched by hand to its invoice, contract or remittance, with the supporting page noted in the workpaper.
- 03
Journal testing
General ledger exports from Xero, MYOB or the client's ERP filtered in spreadsheets to find round amounts, weekend postings and manual journals.
- 04
Workpapers and review notes
Lead schedules and standard workpapers rebuilt every year, and review notes cleared one at a time before the final file is assembled.
Services
All seven services →We take these jobs off your team
Evidence requests
Each client's request list built from last year's file and the audit plan, and chased on a schedule in your words. Every document that arrives is filed against its request in your audit software.
Evidence matching
Source documents read and matched to each sample item, with differences from the ledger listed. The auditor reviews every match and decides what it means for the audit.
- Invoices, contracts, bank confirmations and minutes matched to each sample item, with the page that supports it cited.
- Ask what the prior year file said about a lease or a related party and get the answer with the workpaper behind it.
- Staff see through search only the client files that they can already open in your audit software.
- A private model on your own servers or in an Australian cloud region. No client document is sent to a public AI service.
Sampling support
Populations pulled from the ledger and agreed to the trial balance, then samples selected by the method and size that the auditor sets. The selection and its basis are recorded in the workpaper.
Journal testing
Every journal in the ledger export checked against criteria set by the engagement team, such as round amounts, weekend postings or manual entries to revenue. Each flagged entry is listed with the reason, and the auditor decides which entries to test.
Workpapers and review notes
Lead schedules and standard workpapers drafted from the trial balance and last year's file in your own templates. Open review notes are listed by workpaper and owner before the file is assembled.
SMSF audits
Bank statements, investment reports, member statements and the trust deed read and checked against the fund's financial statements from BGL or Class. Possible contraventions are listed for the auditor, who decides what to report to the ATO.
Register
Audit runs under its own rules
- 01Auditing and Assurance Standards Board ↗
Audit documentation
ASA 230 says an appropriate time limit to assemble the final audit file is ordinarily not more than 60 days after the auditor's report. After assembly, the auditor must not delete or discard audit documentation of any nature before the end of its retention period. Section 307B of the Corporations Act 2001 requires audit working papers to be kept for seven years after the date of the audit report. ASIC can set an earlier date.
- 02Federal Register of Legislation ↗
Auditor rotation
Section 324DA of the Corporations Act 2001 applies to the audit of a listed company, listed registered scheme or registrable superannuation entity. An individual who plays a significant role for 5 successive financial years must then stay out of that role for at least 2 successive years. No individual can play a significant role for more than 5 out of 7 successive financial years. Under section 324DAA, the directors can approve up to 2 more successive years by resolution.
- 03Australian Taxation Office ↗
SMSF annual audit
An SMSF trustee must appoint an approved SMSF auditor each year, no later than 45 days before the SMSF annual return is due. The audit must be complete before the trustee lodges the return, because the return uses information from the audit report. The auditor must be registered with ASIC and independent of the fund. The auditor must report certain contraventions to the ATO, even if the trustee ended the engagement or the audit was not finished.
- 04ASIC ↗
ASIC audit file reviews
ASIC will review 25 audit files in 2026-27, across listed and unlisted companies, registrable superannuation entities and a selection of managed investment schemes. It selects files where a financial report had a material correction or where ASIC is concerned about a material misstatement. It also selects files from data that shows a risk to audit quality, including independence threats, and at random. ASIC will monitor and report on how firms carry out the remedial actions that they committed to.
- 05Office of the Australian Information Commissioner ↗
Personal information in AI tools
OAIC guidance published on 21 October 2024 recommends, as best practice, that organisations do not enter personal information into publicly available generative AI tools. The recommendation applies particularly to sensitive information. The guidance treats information entered into a proprietary AI system as a use rather than a disclosure when protections keep it inside the organisation.
Questions
Where do client files and the model sit?
On your own servers or in an Australian cloud region that you choose. The model runs there too, and no client file, ledger export or workpaper is sent to a public AI service. Staff see through the system only the client files that they can already open in your audit software.
How does this fit the auditing standards?
We build the system to work inside your audit methodology and your system of quality management under ASQM 1. Every match, selection and flag links to its source document, and the workpaper records what the system did and who reviewed it. Whether the evidence is sufficient and appropriate stays a judgement for the engagement team.
Does the AI form the audit opinion?
No. It prepares and indexes the evidence, drafts workpapers and flags items for review. The engagement partner forms the opinion and signs the audit report, and a reviewer signs off each workpaper, as they do today.
We already use Caseware. Do we need to change?
No. What we build reads from and writes back to the audit, ledger and SMSF software that you already run. We will tell you in discovery if a system is the actual constraint.
Where we do this work
Sydney / Melbourne / Brisbane / Perth / Adelaide / Gold Coast / Canberra / Newcastle / Hobart / Darwin
More in financial services
Finance / Banking / Insurance / Real estate / Accounting / Asset and wealth management / Payments and fintech
Before you call
Tell us where your hours go.
We map the work, size the hardware and price what is worth automating.
Talk to our team →