The audit is the judgement — but the engagement runs on writing: the PBC list, the endless "still waiting on…" chases, turning a raw finding into a management point a board will actually read, and the working-paper narratives that eat an afternoon. That drafting is exactly where AI saves time, so you spend more of the day on evidence and less on wording.
The short version: pick a prompt below, paste it into any capable AI assistant (ChatGPT, Copilot, Gemini or Claude — free tiers are fine), swap the [bracketed] details for the engagement's real specifics, read what comes back and edit it before it's used. No new software, no setup.
The boundary matters more here than almost anywhere: AI drafts the words, you perform the audit. It does not gather evidence, run procedures, assess materiality, exercise professional skepticism, or form the opinion — those are yours and can't be handed to a tool. Every prompt is written to keep the judgement with you and treat AI as a drafting aid, nothing more.
01
PBC & document requests
The "prepared by client" list and the requests that follow it are pure logistics — and pure time. A clear, specific request gets answered; a vague one bounces back and forth for a week.
You decide what evidence you need. AI drafts the request from the areas you specify — it doesn't decide scope, sample or sufficiency.
Draft a PBC request list
Turn my list of audit areas into a clear, client-friendly PBC (prepared-by-client) request. Client: [client name], period: [period]. Areas and the items I need for each: [paste your areas and items, e.g. Cash — bank confirmations and reconciliations; Revenue — sales listing and a sample of contracts]. Format it grouped by area, one item per line with a short plain-English note on what it should be, plus a column for status. Add a friendly intro and how/by when to return it. Don't add items I haven't listed.
Tip: let AI format and word it, but keep control of which items appear — the scope is your call, not the tool's.
First reminder — outstanding PBC items
Write a polite first reminder to [client contact] listing the PBC items still outstanding: [paste the outstanding items]. Note that fieldwork is scheduled for [dates] and these are needed to stay on track, make it easy to see exactly what's left, and offer to jump on a quick call if anything's unclear. Warm and organised — this is the first nudge.
Explain why an item is needed
A client is pushing back on providing [the item, e.g. detailed subledger / a sample of contracts], saying [their objection]. Help me write a short, professional reply that explains in plain terms why we need it for the audit, acknowledges their concern, and offers a practical way to provide it ([e.g. read-only access, redacted copies]). Keep it collaborative, not adversarial. Base it only on the reason I give: [your reason].
Confirmation / representation request
Draft a clear cover message to accompany a [bank / legal / receivables confirmation OR management representation] request for [client name]. Explain what we're asking them (or the third party) to confirm, the deadline ([date]), and how to return it. Keep it precise and neutral. Do not draft the substantive content of the confirmation itself — I'll provide that; just the covering communication.
02
Client & engagement communication
Kicking off, keeping people informed, and managing the moments where timelines slip. Clear, calm communication keeps an engagement running — and keeps the relationship intact when something's late on their side.
Engagement kick-off email
Write a professional kick-off email to [client contact] for the [period] audit. Cover: the planned timeline ([key dates: planning, fieldwork, completion]), who's on our team and their roles ([names/roles]), what we'll need from them up front, and a request to confirm the main contacts on their side. Confident and organised, sets a collaborative tone. Under 250 words.
Fieldwork status update
Turn my notes into a short status update for [client contact] partway through fieldwork. My notes: [what's done, what's in progress, what's outstanding, any items slowing us down]. Structure it: where we are, what we still need from them (clearly listed), and whether we're on track for [completion date]. Factual and reassuring, no jargon. Only use what's in my notes.
Flag a timeline slip
Help me tell [client contact] that the completion date may slip because [reason, e.g. several PBC items are still outstanding]. Write a calm, non-blaming message that states the situation plainly, explains the impact, sets out what needs to happen to get back on track and by when, and offers to discuss. Professional and solution-focused. Facts only: [the facts].
Closing / clearance meeting agenda
Draft a concise agenda for a closing meeting with [client / audit committee] for the [period] audit. Topics I want to cover: [list, e.g. status and timeline, significant findings, control observations, required adjustments, next steps]. Format it as a clean agenda with time allocations, lead person per item, and a one-line objective for the meeting. Keep it professional and tight.
03
Findings & management-letter points
Where AI earns its place: turning a finding you've already reached into something a non-finance reader understands, in the structure a management letter expects — without changing what the finding is.
You reach the conclusion; AI writes it up. Give it the finding you've established from your evidence. It structures and clarifies the language — it does not decide whether something is a deficiency or how significant it is.
Structure a finding (condition–criteria–cause–effect)
Write up this audit finding in the standard structure — condition, criteria, cause, effect, and recommendation. My raw finding: [what you observed, against what should be, why it happened, the impact, and your recommended action]. Keep it factual and objective, avoid overstated language, and make the recommendation practical. Use only what I've given you — don't infer causes or effects I haven't stated.
Explain a control deficiency to a non-expert
Explain this control deficiency in plain English to a board member who is not an accountant. The deficiency: [describe it]. The risk it creates: [the risk, in your words]. Write 2–3 short paragraphs: what the weakness is, why it matters in everyday terms, and what fixing it would involve. Measured and clear, not alarmist. Only what I've provided.
Soften the tone without softening the finding
Reword this management-letter point so the tone is professional and constructive while the substance stays exactly the same. My draft: [paste it]. Keep every fact, the severity and the recommendation unchanged — only adjust wording that reads as accusatory or vague. Do not downgrade or soften what the finding actually says.
Summarise findings for the executive summary
Turn my list of findings into a short executive summary for the front of the management letter. Findings (with the rating I've assigned each): [paste them]. Give a 3–4 sentence overview, then a brief line per finding grouped by significance, in the order I've rated them. Keep it balanced and factual, don't re-rate anything, and don't add findings that aren't on my list.
04
Working-paper write-ups
The documentation that takes as long as the testing. Feed AI the notes you took while doing the work and it drafts a clean narrative — which you then review, correct and own.
Narrative from testing notes
Turn my rough testing notes into a clear working-paper narrative. Area: [e.g. revenue cut-off testing]. My notes: [what you tested, the population and sample, what you did, what you found, your conclusion]. Structure it: objective, work performed, results, and conclusion. Keep it factual and in the past tense, use my figures exactly, and don't state a conclusion beyond the one I've written.
Tip: AI documents what you did — it can't be the evidence. Keep your tickmarks, samples and conclusions; let it tidy the prose.
Document a process walkthrough
Write up a walkthrough narrative from my notes. Process: [e.g. purchases-to-pay]. What I observed step by step, including the controls in place and who performs them: [paste your notes]. Lay it out as a clear step-by-step description with the key control identified at each relevant step. Factual and neutral; don't assess whether controls are effective — just describe what I observed.
Summarise an interview / meeting for the file
Summarise this discussion with [name/role] for the audit file. My notes: [paste them]. Give: attendees and date, the purpose, the key points made, and any items to follow up with who owns each. Keep it to what was actually said — don't add interpretation or conclusions I haven't recorded.
Memo explaining a judgement
Help me write up a memo documenting a judgement I've made. The judgement and my reasoning: [state the judgement, the factors you considered, and your conclusion]. Structure it: the matter, the considerations, and the conclusion reached. Present my reasoning clearly and completely — the judgement is mine; your job is only to document it faithfully, adding nothing.
05
Fieldwork queries
The quick questions that come up mid-testing — an unusual item, a variance, a reconciling difference. A crisp query gets a fast answer and keeps fieldwork moving.
Query an unusual item or variance
Write a short, neutral query to [client contact] about something I've noticed while testing. What I see: [the item/variance, factually — e.g. [account] increased [amount] on [date] with no obvious explanation]. Ask them to explain what drove it and provide any supporting document, without implying anything is wrong. One short paragraph, professional and specific.
Ask about a reconciling difference
Help me query a reconciling difference with [client contact]. Details: [what reconciles to what, and the difference — amount, accounts, period]. Ask them to identify and explain the difference and provide support, and offer the 2–3 most likely explanations I've considered so they can confirm quickly. Neutral and precise.
Chase an unanswered query
Write a courteous follow-up on a query [client contact] hasn't answered yet. Original question: [restate it]. Sent on [date]. Note briefly why it's holding up progress on [area], keep it friendly, and make it easy to reply. Short.
Confidentiality and judgement, in one line. Audit-client information is confidential — check what any AI tool does with your data, keep sensitive figures and identifiers out of your prompts, and remember that the evidence, the skepticism and the opinion are yours. AI drafts the words; it never audits.
06
Common questions
Which AI tool should an auditor use?
Any capable general assistant works for the drafting on this page — ChatGPT, Microsoft Copilot, Google Gemini or Claude. Free tiers are enough to try every prompt. The point isn't the brand but the boundary: it drafts the words around your work, it doesn't perform the audit.
Can AI do the audit or gather evidence?
No. A general AI assistant is a writing aid, not audit evidence. It doesn't run procedures, test balances, assess materiality, exercise professional skepticism or form the opinion — those can't be delegated to a tool. Use it to draft the PBC list, the status email or the plain-English management point, then apply your own judgement to everything it produces.
Does using AI affect my independence?
Used as a drafting aid for your own communications and write-ups, a general assistant is like any word processor — it doesn't create a self-review or advocacy threat on its own. What matters is that the audit work, the conclusions and the judgements stay entirely yours, and that you follow your firm's policies on tools and data.
Is it safe to paste audit-client information into an AI tool?
Audit-client data is confidential and often sensitive. Treat any AI tool as an external service, follow your firm's confidentiality and data policies, and keep specific figures, names and identifiers out of the prompt where you can — use the [bracketed] placeholders and give only the context the draft needs. Check what the tool does with your data first.
If you'd rather the admin ran itself
These prompts are the manual version — copy, paste, edit, use. They work, and for a lot of firms that's exactly enough. But if the same PBC-chasing, status-writing and documentation happens on every engagement, that's the part that can run quietly inside the tools your firm already uses, with client data handled properly and a person always reviewing before anything goes out.
That's what SG1 Consulting builds with The Everything — and there's a free 22-page playbook that shows exactly how a practice puts it to work.