AI-assisted audit execution

Audit testing,
automated.

Less ticking. More thinking.

Sample, vouch, evaluate, document — the substantive testing that eats your busy season, run end to end in under an hour instead of fifteen. The same procedures, the same standards. Your judgement, not your time.

Book a demo → Try it live

✓ Time compression demonstrated in testing on live-shaped engagement data.

Per revenue engagement

~15hours
by hand

→

under 1hour
with Kritos

✓End to end — sample to signed workpaper, in one run
✓Built to the standards — ASA 530 sampling, ASA 450 evaluation
✓Built by a practising auditor — every step reflects real audit judgement

≈ 1,400 expert hours — about $350k in billable capacity — back across a 100‑engagement year · illustrative

ASA 530 / 450 methodology Hosted in Australia · Azure No engagement content retained Built by a Chartered Accountant

The last manual frontier

Everything around the testing got automated — scheduling, confirmations, the file. The substantive work itself, tracing a sample to source, computing variances, evaluating misstatement, is still done by hand, every busy season. Kritos automates that. Defensibly.

10–20hoursof manual work to verify one company’s reported revenue — on a single audit, every year.

Chargeable hours, underutilised. Senior time is absorbed by repetitive testing — not the analysis and client insight that defines a quality engagement.
The work is scattered. Auditors move between spreadsheets, extraction software and manual checks — re‑keying data and stitching steps at every stage.
Graduates learn ticking, not judgement. Juniors spend their early years on repetitive procedures instead of building the judgement that accelerates a career.

Try it — no login

Substantive testing, live.

Set your performance materiality and run the procedure. Kritos derives the MUS sample, reads and matches each document, evaluates the variances, and projects the misstatement across the population. Where it flags something, amend it yourself — the variance and the conclusion recompute live. You stay in control.

$14.2M

// revenue population · 240 transactions

Performance materiality $240,000

$200k$600k

$80,000

Sampling interval

113

MUS sample

Target tested · 100%

Sample derived per ASA 530 (confidence factor 3.0). Clearly-trivial threshold $5,000 — an independent auditor input, never a % of materiality.

Workpaper — Substantive Testing Ready

#GL (ex GST)InvoiceVarianceResult

Set materiality, then run the procedure.

What Kritos does

From population to workpaper, in one pass.

Not a chatbot bolted onto a spreadsheet. A testing engine that performs the procedure end to end. Step through it:

ASA 530 · Audit Sampling

Monetary-unit sampling, derived in full

Confidence factors, risk assessment, target testing, the residual population — the complete derivation, documented so a reviewer can re-perform it. The sample responds to materiality, exactly as the standard intends. Not a black box.

selected at interval

$850,000

$425,000

$680,000

$312,400

interval $93,333 · high-value items 100% examined

Occurrence · Accuracy

Every item vouched to its document

Each sampled transaction reconciled to its supporting invoice — number, amount, date, currency — with the evidence highlighted on the page. Foreign-currency invoices are handled deterministically: where the document states its own AUD, that figure is used; the auditor can correct any extraction.

GL · INV-PFZ-2025-003$312,400

⇅ matched on number + amount

Invoice · INV-PFZ-2025-003$309,950

✓ evidence captured · original preserved

ASA 450 · Evaluation of Misstatements

Variance computed and extrapolated

Differences classified against clearly-trivial and performance-materiality thresholds, then projected over the residual population. High-value items examined in full are taken as actual, the remainder projected — every number ties to the next.

✓ ≤ clearly-trivial → Pass

! > CT, < PM → Unadjusted difference

■ ≥ PM → Material — investigate

Projected misstatement$215,000

Audit-ready output

Review-ready, not a starting point

Sampling derivation, testing results, exceptions, extrapolation — assembled into a workpaper your reviewer can sign. Nothing of the engagement data is retained afterward. The auditor reviews and concludes; Kritos did the tracing.

Workpaper — 9 sheets

Audit parameters✓

MUS sample · testing results✓

Variance extrapolation✓

✓ exported · no content retained

Your firm’s numbers

What it’s worth to your firm.

Set the dials to your firm. The estimate applies the ~15‑hours‑to‑under‑an‑hour compression demonstrated in testing — to the revenue testing you already do.

Revenue-testing hours / year1,500

2006,000

Charge-out rate$250 / hr

$100$600

Employee cost$65 / hr

$30$150

$350,000

billable capacity freed each yearhours back × charge-out rate

1,400

expert hours back — for advisory, analysis and more engagements

$91,000

staff cost absorbed by repetitive testing todayhours back × employee cost

Illustrative — built on the time compression demonstrated in testing, not a measured pilot result. Your numbers depend on your engagements and rates.

Why existing tools fall short

Plenty of tools. None go end to end.

Firms already own pieces of this. Each tool automates one step, then hands the work back — so the job gets stitched together by hand across several of them. Kritos runs all of it as a single workflow.

Tool categoryWhat it doesWhy firms stop short

Analytics & sampling suites

Sampling and data analytics

Complex, costly per seat, heavy training — too cumbersome for everyday testing.

PDF extraction tools

Pull document data into Excel

Begin at extraction — no sampling, no methodology, no matching.

In‑Excel matching add‑ins

Match documents inside Excel

Excel‑dependent; no sampling engine, no methodology layer.

Manual Excel (MUS)

Auditor‑built sampling model

No AI, manual entry throughout, no document processing, high error risk.

Kritos

Sampling → extraction → AI matching → workpaper

One workflow, end to end — built to ASA 530 and ASA 450.

No single tool combines sampling, extraction, AI matching and the workpaper into one job — so firms run several in parallel and re‑key between them. Kritos closes that gap.

Where this goes

The audit file that assembles itself.

The destination is a substantive-testing layer that spans the whole file — leaving the auditor to the judgement, not the tracing. Two cycles run today; the rest are on the way. Kritos is building it for the firms that get there first.

RevenueLive

ExpensesLive

Trade debtorsNext

PrepaymentsNext

PayrollNext

Trade creditorsPlanned

AccrualsPlanned

Fixed assetsPlanned

Inventory & COGSPlanned

CashPlanned

BorrowingsPlanned

EquityPlanned

Live today In development On the roadmap

Questions auditors ask

Straight answers.

No. Kritos performs the mechanical work — sampling, vouching, computing variances — and drafts a conclusion. You make every determination, amend anything you disagree with, and sign off. The judgment, and the conclusion, stay yours.

Yes. Monetary-unit sampling follows ASA 530, and misstatement evaluation and extrapolation follow ASA 450 — throughout. It is not a US framework relabelled for the local market.

Engagement data is hosted on Microsoft Azure, Australia East. Row-level security isolates each firm, so your data is never visible to another tenant. The AI layer retains nothing beyond the API request, and uploaded files are removed after the workpaper is exported.

No — and it matters to be clear about that. Kritos tests occurrence and accuracy by vouching recorded items to their source documents. Completeness — the search for unrecorded items — remains your separate procedure. Kritos doesn’t claim to do everything.

You are. Kritos assembles the evidence and a draft conclusion; the auditor reviews it, applies judgment, and signs. Kritos is a tool that does the tracing — it is not a signer.

There isn’t one hiding in the methodology — it’s the same procedures, traced and documented faster, and demonstrated in testing on live-shaped engagement data.

The founder

Built by Ishwar Chugh, CA.

Kritos came out of the work itself. For years, every busy season meant doing substantive testing by hand — drawing the sample, chasing documents, tracing figures, assembling the workpaper — while everything around it had already been automated. So I built the part that hadn’t: the substantive testing itself. It’s live at kritos.com.au, running real audit data end to end, with the time savings demonstrated in testing and pilot discussions underway.

Audit manager based in Melbourne. Chartered Accountant (CAANZ), Master of Management in Accounting and Finance from the University of Melbourne. Several years of statutory audit practice across two Australian firms — graduate through manager level.

“Every workflow decision in Kritos was made by someone who has done the work, supervised it, and signed it off.”

How a pilot works

From first look to live workpaper.

Low commitment, high signal — see it on real work before you decide anything.

1STEP 01

See it live

A short walkthrough on representative engagement data — sampling to workpaper, end to end. No setup required.

2STEP 02

Run a pilot

Kritos on one of your real engagements — your data, your materiality, your review. Supported the whole way.

3STEP 03

Keep the workpaper

A review-ready workpaper out the other side. Your conclusion, your sign-off. Judge it on the output.

See it on your own work

Bring a sample.
Watch it run.

A short demo on real engagement data — sampling to workpaper, end to end. See exactly what your reviewer would see.

Book a demo → Sign in →

Audit testing,automated.