All features/Working papers & review notes

Phase 7 · Execution

Practical working papers — not tick-box checklists.

Execution is where Audit Intelligence breaks from the tick-box tools: working papers are structured digital objects — lead sheet, sampling, procedures, evidence, conclusion — generated from your mapping and risk matrix, with AI doing the grunt work and the auditor signing every conclusion.

Structure

Working papers as data, not documents.

Every working paper is a structured object: sections and questions, AI suggestions with source citations, auditor responses, status, and linked evidence. They are auto-generated per financial-statement area from the FSA mapping and the risk matrix, and every one runs the same shape — so a graduate who has done one revenue file can navigate any file.

  • Colour coding everywhere: blue headings are client data, green headings are procedures AuditBot can perform.
  • AuditBot greets the first open with a plain-English summary of what the working paper is for.
  • TLDRs top complex sections; progress is tracked per question — 95% shows as 95%.
app.audit-intel.com/engagements/acme-fy2025/wp/revenue
Lead sheetSampling (MUS)ProceduresEvidenceConclusion
Blue = client-preparedGreen = AI-performable
Lifecycle

A sign-off chain that mirrors the firm.

Working papers move Not started → In progress → Ready for review → Reviewed → Notes addressed → Completed. Preparers prepare, reviewers review or raise notes, and the partner is mandatory on the gates that matter: acceptance, independence, materiality, the risk register and matrix, the opinion, and file lock.

  • Every status change is logged with who and when.
  • Print and export views produce clean, reviewable output per working paper.
app.audit-intel.com/engagements/acme-fy2025/working-papers
Not startedIn progressReady for reviewReviewedNotes addressedCompleted
A working paper cannot be completed while it has open review notes — a hard gate, not a convention.
Sampling

MUS sampling with the discipline built in.

Monetary Unit Sampling runs at transaction level. The population must reconcile to the lead sheet within performance materiality before sampling proceeds. Exclusions apply first — material items go to 100% testing, interim-tested items, journals (tested separately), and prior coverage come out — then sample size is calculated from the remaining population, tolerable misstatement and a 95% confidence level, with an AI-generated unbiased random start and systematic selection. Small populations route to a high-value-items approach per ASA 530 instead.

  • The random start is AI-generated between 1 and the interval — unbiased and auditor-verifiable.
  • Every parameter and selection is on the file for re-performance.
app.audit-intel.com/engagements/acme-fy2025/sampling
Population reconciled to lead sheetWithin PM
Material items (100% tested)8 items · $1.2m
Journals excludedTested via fraud analytics
Sample selected27 items · random start verified
Journals & SUAD

Journals that flow — adjusted and unadjusted.

Adjusting journals post through to lead sheets and the draft financial statements in real time, with a summary report for the client to post in their own system. Journals the client declines to correct flow automatically to the Summary of Unadjusted Audit Differences — no manual collation — aggregated for the materiality comparison at completion.

  • Journal types: adjusting, reclassifying, eliminating, unrecorded (factual / projected / judgemental).
  • All journals are excluded from standard sampling and tested separately for fraud characteristics.
app.audit-intel.com/engagements/acme-fy2025/journals
AJE-3 · Revenue cut-offPosted → lead sheet updated
AJE-5 · Inventory NRVClient declined → SUAD
SUAD vs overall materialityClear by $41,300
Fraud analytics

Twenty-five tests across the whole journal population.

Journal-entry testing per ASA 240 starts with population completeness, then AuditBot runs a 25-test analytics framework across six layers — timing, user and role, source, amount, account and description, and aggregate patterns. Each journal gets a risk score from its flags, building an outlier register: entries meeting pre-set criteria are 100% tested, with MUS applied to the flagged remainder.

  • Round-dollar, threshold-hugging, after-hours, same-user-creates-and-approves — the classic red flags, run systematically.
  • Vouching covers authorisation, business rationale and accounting treatment, with conclusions signed off.
app.audit-intel.com/engagements/acme-fy2025/fraud-analytics
TimingUser / roleSourceAmountAccount / descriptionAggregate
Journals analysed4,182
Flagged — outlier register61
Posted on weekends by management7 — 100% tested
Review notes

A separate portal — because inspection files get cleaned.

Review notes live in their own portal, deliberately off the main audit file, preserving the quality-management record without parking commentary where a regulator reads. Every note is linked to its working paper or lead sheet — no orphan notes — and moves Raised → Resolved → Closed. Files flag working papers with open notes, dashboards count them per person, and completion is blocked while notes stay open.

  • Notes carry subject, type, priority, assignee and a conversation thread.
  • Everyone on the file sees all notes on that file — nothing happens in side channels.
app.audit-intel.com/engagements/acme-fy2025/review-notes
Revenue cut-off — clarify sample 14Raised
Debtors — add subsequent receiptsResolved
Cash — agree to bank letterClosed
Hard gate: a working paper cannot be signed off as complete with active review notes.

See it on your own engagements.

Start your firm free, or book a demo and we will walk you through the platform on a real audit file.

No credit card to start · Australian-resident data · Cancel anytime