Act as a forensic-audit analyst helping me organize and test the records below. This is an investigative workpaper, not a legal conclusion, accusation, or replacement for a licensed forensic accountant, auditor, or attorney. Audit objective: [AUDIT OBJECTIVE] Entity and period: [ENTITY AND PERIOD] Available records: [RECORDS AND DATA] Known concerns or red flags: [RED FLAGS] Materiality threshold or amount at risk: [MATERIALITY THRESHOLD] Relevant policies, approvals, or expected process: [CONTROL EXPECTATIONS]
Use only the information I supply. Do not infer fraud, identify a culprit, alter source figures, or fill gaps with invented explanations. Produce a forensic-audit workpaper with these sections: 1. Scope and limitations: state the objective, period, sources reviewed, missing records, and assumptions. 2. Evidence inventory: list each source, its date range, owner, reliability concerns, and what it can substantiate. 3. Transaction testing plan: define population, selection logic, fields to compare, duplicate-payment tests, round-dollar and threshold tests, split-transaction tests, vendor/employee conflict checks, timing cut-off checks, approval-chain checks, and bank-to-ledger reconciliation steps that fit my facts. 4. Findings table: for every exception, show record ID, date, amount, source evidence, calculation, relevant policy/control, severity, alternative explanations, and follow-up evidence needed. 5. Reconciliations: show each calculation line by line, including opening balance, additions, deductions, and unresolved difference. 6. Recommended next steps: preserve evidence, request records, expand samples, and identify issues requiring an independent professional. Before answering, check that every number can be traced to an input and flag any unsupported claim. Also distinguish a confirmed discrepancy from a red flag or an untested hypothesis. Ask up to 3 clarifying questions only if a required input is missing.
Fill in
| Placeholder | What to enter | Example |
|---|---|---|
| [AUDIT OBJECTIVE] | State the question the audit must answer, such as whether duplicate vendor payments occurred. | Determine whether marketing vendors were paid twice or without the required approvals. |
| [ENTITY AND PERIOD] | Enter the organization or account name and the date range under review. | Northfield Foods LLC, January 1 to June 30, 2026. |
| [RECORDS AND DATA] | Paste or summarize the ledger exports, invoices, bank statements, emails, and other available records. | AP ledger CSV, 84 invoices, May and June bank statements, vendor master export, and approval emails for 22 payments. |
| [RED FLAGS] | List observed anomalies, complaints, or transactions that prompted the review. | Three payments just below the $10,000 approval limit and two invoices sharing an amount and vendor name. |
| [MATERIALITY THRESHOLD] | Enter the amount or rule for deciding which exceptions require escalation. | Escalate exceptions over $2,500 or any payment lacking two required approvals. |
| [CONTROL EXPECTATIONS] | Describe the required approval, purchasing, payment, and recordkeeping process. | Purchases over $10,000 need department-head and CFO approval; AP matches PO, receiving record, and invoice before payment. |
How to use
- Gather the original exports and records, keeping transaction IDs and dates intact.
- Paste the prompt and replace every bracketed field with your audit scope and evidence.
- Check the findings table against source documents and recalculate each unresolved difference.
- Follow up with: “Turn the findings into a records-request list prioritized by evidentiary value and preservation risk.”
Variations
Vendor payment review
Use this when the concern is limited to accounts payable and vendor billing.
Review [AP LEDGER], [INVOICES], [BANK PAYMENTS], and [APPROVAL POLICY] for duplicate, altered, unauthorized, or split vendor payments from [PERIOD]. Create a test matrix matching invoice number, vendor tax ID, amount, date, PO, receiving evidence, approver, and bank reference. Return exceptions with source IDs, exact calculations, plausible non-fraud explanations, and documents needed to resolve each one. Do not allege fraud or invent missing records. Flag conclusions that need forensic-accounting or legal review.
Expense claim review
Use this when reviewing employee reimbursements or corporate-card activity.
Analyze [EXPENSE DATA] for [PERIOD] against [EXPENSE POLICY] and [TRAVEL CONTEXT]. Test duplicate receipts, weekend or personal-looking charges, missing business purpose, policy-limit breaches, mileage anomalies, split claims, and reimbursements also paid by card. Produce an exception register with employee, transaction ID, amount, policy reference, evidence, risk level, and a neutral follow-up question. Use only supplied evidence; distinguish errors from unverified concerns.
Cash reconciliation
Use this when cash, deposits, or petty-cash balances do not reconcile.
Reconcile [GENERAL LEDGER], [BANK STATEMENTS], [DEPOSIT LOG], and [CASH COUNT RECORDS] for [PERIOD]. Build a dated bridge from book balance to bank balance, separately identify outstanding checks, deposits in transit, fees, reversals, and unexplained items. Show every arithmetic step and retain source references. Then list cash-control weaknesses and evidence requests. Do not assume theft from an unexplained difference; recommend professional escalation for material or persistent discrepancies.
Tips
- Preserve original exports and files before sorting, filtering, or annotating them; your audit trail should retain dates, IDs, and source locations.
- Test the complete transaction population for duplicates and threshold patterns before selecting judgmental samples.
- Treat a policy breach and a proven financial loss as separate findings; one does not automatically establish the other.
- Keep privileged legal communications separate from ordinary audit workpapers if counsel directs the investigation.
FAQ
Can AI determine that fraud occurred?
No. It can organize evidence and identify anomalies, but fraud findings require corroborated evidence and often forensic-accounting and legal review.
What records are most useful for a payment investigation?
Start with the ledger, bank detail, invoices, purchase orders, receiving records, vendor master data, approval records, and relevant communications.
Should I upload sensitive financial records?
Use your organization’s approved AI environment, minimize personal data, and follow retention, confidentiality, and legal-hold requirements.