Create a copy-paste AI prompt pack for a chartered accountant working in [PRACTICE AREA] in [JURISDICTION]. The recurring work I want covered is [RECURRING TASKS]. My source materials and systems are [SOURCE MATERIALS]. The intended users and review level are [USERS AND REVIEW PROCESS]. Build 8 distinct prompts, each for a task that an accountant would actually perform. Prioritize the listed recurring work and include only tasks appropriate for the stated practice area. Do not provide legal, tax, audit, or accounting conclusions as settled facts when authority, client facts, or professional review is required. Do not invent ledger entries, source documents, standards citations, client facts, or calculations. For each prompt, provide: - A short task title and the best moment to use it. - A copy-paste prompt of 120-180 words, written as the accountant speaking to an AI. - Clearly labeled square-bracket inputs. - An exact output format, such as a reconciliation exception table, workpaper outline, issue log, client-information request, memo structure, or review checklist. - A built-in self-check requiring the AI to distinguish source evidence from inference, flag missing evidence, and identify matters for partner, manager, tax specialist, or audit-signing review. Cover a balanced set where relevant: transaction classification or reconciliation, variance analysis, audit planning or testing support, control walkthroughs, technical research questions, client information requests, management reporting, and review notes. Include the applicable reporting or tax framework only when supplied, and instruct the AI to quote or cite only material the user provides or independently verifies. Finish with a one-page prompt selection guide mapping common situations to the right prompt. Ask up to 3 clarifying questions only if the practice area, jurisdiction, recurring tasks, or source material is missing.
Fill in
| Placeholder | What to enter | Example |
|---|---|---|
| [PRACTICE AREA] | State the accounting role or service line, such as audit, tax, advisory, or financial reporting. | External audit for owner-managed companies. |
| [JURISDICTION] | Name the country and any relevant local regulatory context. | England and Wales, with UK GAAP reporting engagements. |
| [RECURRING TASKS] | List the day-to-day tasks the prompt pack should address. | Planning materiality, bank reconciliations, revenue testing, control walkthrough notes, audit queries, and manager review responses. |
| [SOURCE MATERIALS] | Describe the ledgers, reports, contracts, standards, and client documents available. | Trial balances, bank statements, sales invoices, contracts, prior-year files, client explanations, and firm audit templates. |
| [USERS AND REVIEW PROCESS] | State who will use the prompts and who must review the resulting work. | Audit seniors will use the prompts; managers review workpapers and the engagement partner signs the audit opinion. |
How to use
- List your real recurring tasks and the framework you work under; do not ask one generic prompt to cover every accounting service.
- Paste a representative workpaper template or review-note style if your firm has one.
- Check each generated prompt requires source evidence, documents uncertainty, and routes judgments to the appropriate reviewer.
- Follow up with: "Rewrite prompts 2, 4, and 6 to match this workpaper template and keep every evidence reference traceable."
Variations
Audit planning pack
Use when preparing prompts specifically for audit planning and fieldwork.
Create six copy-paste prompts for an audit team planning [CLIENT TYPE] under [REPORTING FRAMEWORK] in [JURISDICTION]. Available material includes [SOURCE DOCUMENTS]. Cover risk assessment, materiality documentation, revenue risk, controls walkthroughs, sample-selection rationale, and audit-query drafting. Each prompt must request a defined workpaper output, distinguish client assertions from audit evidence, and flag judgments for manager or partner review. Do not invent standards requirements, audit evidence, or test results. Add a short use note for each prompt.
Month-end close pack
Use when supporting an in-house finance close process.
Create seven AI prompts for a [COMPANY TYPE] finance team completing month-end close in [ACCOUNTING FRAMEWORK]. Systems and reports: [SYSTEMS]. Close problems to address: [CLOSE ISSUES]. For each prompt, include inputs, a required output table or checklist, reconciliations or cutoff checks, and an escalation rule for unsupported entries or material variances. Cover close calendar review, balance-sheet reconciliations, accrual support, variance explanations, flux analysis, journal-entry review, and management reporting. Never fabricate balances or approve entries.
Tax client queries
Use when organizing tax information requests and research work.
Draft five copy-paste prompts for a tax adviser in [JURISDICTION] handling [CLIENT TYPE] questions about [TAX TOPICS]. Use only these possible sources: [SOURCES]. Include prompts for fact gathering, issue spotting, research plan creation, client-information requests, and a technical memo outline. Each must separate facts, assumptions, and items requiring verification against current authority. State that the output supports qualified professional review and must not invent tax law, dates, elections, or client transactions.
Tips
- Use AI to organize evidence and expose gaps, but preserve the source-document trail that lets another accountant reperform the work.
- For reconciliations, require an exception table with opening balance, movements, closing balance, source reference, timing difference, owner, and resolution status.
- For audit work, distinguish the client's explanation from corroborating evidence; a clear narrative is not audit evidence by itself.
- For technical matters, ask for a research plan and authority checklist before asking for a conclusion, especially where jurisdiction or effective date may matter.
FAQ
Can I use AI to draft audit workpapers?
Yes, for structure, summaries, query drafts, and issue logs, subject to firm policy and review. The workpaper still needs traceable evidence and professional judgment from the engagement team.
How can I prevent fabricated citations?
Tell the model to cite only text you provide or sources it can verify, and require it to mark any unverified authority. Check citations against current primary materials before relying on them.
Should client data be pasted into a public AI tool?
Follow your firm's confidentiality, data-processing, and approved-tool policies first. Remove unnecessary identifiers and use approved enterprise controls where required.