The Thomson Reuters Institute's 2025 Future of Professionals report estimated that AI saves the average CPA around 240 hours a year, roughly 19,000ofrecoveredcapacityperprofessional,anda12 billion gain across the profession [1]. Adoption is no longer fringe: in a Thomson Reuters survey of tax firms, 44% of generative-AI users reported using it daily and another 29% at least weekly [2]. The firms pulling ahead are not the ones with the most expensive AI subscription. They are the ones with a reusable prompt library tuned to the work a tax and accounting practice actually does.
This article collects 50 ready-to-use prompts for accountants and tax advisors, organized by workflow: tax preparation and filings, tax advisory and planning, financial statements and bookkeeping, audit and due diligence, client communication, compliance, practice management, and professional development. Each prompt uses [bracket placeholders] so you can adapt it to your engagement. They work with ChatGPT, Claude, and Gemini, with notes where one model is clearly better for a task.
Accounting workflow map: eight categories where prompts compound the hours you save
Before you paste: two non-negotiable rules
1. Confidentiality and client data. Consumer AI chatbots (the free tier of ChatGPT, standard Gemini, Claude.ai) retain prompts and may use them for training. Pasting a client's tax return, payroll file, bank statements, or any identifying financial data breaches your confidentiality duty and, in many jurisdictions, data-protection law. For real client work, use an enterprise-grade deployment with a contractual no-training commitment and SOC 2 / ISO 27001 controls: ChatGPT Enterprise, Claude for Work, Microsoft Copilot, or a self-hosted option. When in doubt, redact names, identifiers, and account numbers before you prompt.
2. Tax law changes and AI gets it wrong. Models fabricate code sections, thresholds, rates, and filing deadlines with complete confidence, and they are routinely trained on rules that have since changed. Every figure, citation, and deadline an AI returns must be verified against the official source: the tax authority, current legislation, or your professional research platform. Treat AI output as a draft and a thinking aid, never as authority.
Verification decision tree: what you can paste, what needs redaction, and what must be checked against an official source
With those two guardrails in place, here are the 50 prompts.
1. Tax Preparation & Filings
AI is fastest on the first pass: organizing source data, surfacing missing items, and drafting explanations a professional then verifies. Never let it file or finalize.
1. Build a document request list for an engagement
I am preparing the [tax year] return for a [individual / small business / partnership / corporation] with these characteristics: [describe income sources, entity type, prior-year items]. Produce a complete client document request checklist, grouped by category, flagging the items most often forgotten for this client profile. Note which items are needed only if a condition applies, and state the condition.
2. Reconcile a figure across documents
Here are figures for [item, e.g. total revenue] from [N] sources: [paste bookkeeping total, bank summary, prior return, client statement]. Identify every discrepancy, quantify it, and list the most likely reasons for each gap ranked by probability. Tell me which source is normally the most reliable for this item and what to ask the client to resolve it.
3. Explain a notice or assessment in plain language
Summarize this [tax authority notice / assessment] in plain language. State: (1) what the authority is claiming, (2) the amount and period at issue, (3) the deadline to respond, (4) the options available, (5) what information I need from the client to respond. Do not advise on the merits yet; just decode it. Notice: [paste redacted text].
4. Draft a workpaper narrative
Write a clear workpaper narrative documenting how I arrived at [figure / treatment] for [client, period]. Include the source data used, the calculation steps, the rule or policy applied [I will verify the citation], and any judgment calls with their rationale. Write it so a reviewer or future-me can follow it cold. Inputs: [paste].
5. Categorize a transaction list
Categorize these transactions into [chart-of-accounts categories / deductible vs non-deductible / capital vs revenue]. For each, give the category and a one-line reason. Start conservative; flag anything ambiguous or that looks personal-in-business for my review rather than guessing. Transactions: [paste redacted list].
6. Draft a return-review checklist
Produce a pre-filing review checklist for a [return type] for a [client profile]. Group by: data completeness, internal consistency, common error points for this return type, items that frequently trigger a query from the authority, and sign-off steps. Make each item a yes/no I can tick.
7. Spot missing deductions or credits to investigate
Given this client profile [describe activity, expenses, life events, entity type], list deductions, allowances, or credits that a thorough preparer would at least investigate for this profile. For each, note the condition that would make it apply and the documentation I would need. I will verify current eligibility and amounts against the official source.
8. Turn raw notes into a filing-status memo
Convert these intake notes into a short internal memo recommending a filing approach for [client]: [paste notes]. Structure: facts, the question, the realistic options with the trade-off of each, a recommendation, and the open items I must confirm before proceeding. Flag anything where the answer depends on a rule I should re-check.
2. Tax Advisory & Planning
Advisory is where the margin is. AI structures the analysis and drafts the explanation; the judgment and the numbers stay with you.
Want to know how effective your prompts are? Prompt Score analyzes them on 6 criteria.
A client is deciding between [Option A] and [Option B] for [transaction / entity choice / compensation structure]. Their situation: [describe]. Build a side-by-side comparison covering: cash-tax impact over [N] years, timing, administrative burden, flexibility, exit consequences, and risk of challenge. End with the questions whose answers would change the recommendation. I will confirm all rates and rules.
Gemini and Claude both handle multi-column comparison tables cleanly; Claude tends to be better at carrying a long fact pattern without dropping a detail.
10. Model the sensitivity of an estimate
Here is my projection for [client's tax position] under these assumptions: [list]. Show how the outcome changes if each key assumption moves by [+/- range]. Identify the two assumptions the result is most sensitive to, so I know where to focus the client conversation. Numbers: [paste].
11. Draft a year-end planning checklist
Produce a year-end planning checklist for a [client profile, e.g. owner-managed business]. Group actions by deadline-driven, discretionary-but-valuable, and watch-list-for-next-year. For each action, one line on the benefit and one line on the catch. I will verify current thresholds.
12. Pressure-test a planning idea
I am considering recommending [planning approach] for a client whose facts are [describe]. Play the role of a skeptical reviewer and a tax authority examiner. List the strongest objections to this approach, the facts an examiner would probe, the documentation that would defend it, and the point at which this stops being planning and starts being aggressive.
13. Explain a planning trade-off to a non-expert owner
Explain the trade-off between [Option A] and [Option B] to a business owner with no tax background. Use one analogy from running a business. Keep it under 300 words. End with: the upside, the cost or risk, and a clear recommendation framed as "if your priority is X, choose Y."
14. Build a forecast-to-cash-tax bridge
Turn this financial forecast [paste P&L projection] into a rough cash-tax timeline for [period]: when liabilities arise, when payments are due, and where cash-flow pinch points appear. Flag every figure that depends on a rate or rule I need to confirm, and keep them in a separate "verify" list.
15. Draft an advisory engagement scope
Draft a scope-of-work paragraph for an advisory engagement on [topic] for [client]. Be explicit about what is included, what is excluded, the assumptions, the client's responsibilities, and the limits of reliance. Plain English in the body; precise on the exclusions and the reliance limitation.
3. Financial Statements & Bookkeeping
These prompts triage volume and draft the words around the numbers. They do not replace your review of the numbers themselves.
16. Draft notes to the financial statements
Draft the [specific note, e.g. revenue recognition / related-party / going-concern consideration] for the financial statements of [entity type] under [reporting framework, e.g. IFRS / local GAAP]. Use these facts: [paste]. Keep it compliant in structure but readable. Flag any disclosure where the facts I gave you are insufficient to complete it properly.
17. Write a management-accounts commentary
Write the commentary for this month's management accounts: [paste P&L and key variances]. Structure: headline performance in two sentences, the three variances that matter and why, cash position, and one forward-looking flag. Tone: direct, for a busy owner. Under 350 words. Do not invent explanations for variances; where you do not know the cause, say "cause to confirm."
18. Explain a variance
Revenue / [line item] moved [amount or %] versus [prior period / budget]. Here is the underlying detail: [paste]. List the plausible drivers ranked by how much of the movement each likely explains, and tell me what to check to confirm the real cause.
19. Build a bookkeeping cleanup plan
A new client's books are in this state: [describe issues, e.g. uncategorized transactions, mixed personal and business, unreconciled accounts]. Produce a prioritized cleanup plan: what to fix first, the order that minimizes rework, the risky areas to flag to the client, and an estimate of the effort in stages.
20. Map two charts of accounts
I am migrating a client from [system A] to [system B]. Here are the two charts of accounts: [paste]. Produce a mapping table from old to new, flag accounts with no clean equivalent, and propose how to handle each. Note where merging accounts would lose useful history.
21. Draft a close checklist
Build a month-end close checklist for [business type] using [accounting system]. Group by: pre-close prep, reconciliations, accruals and adjustments, review, and reporting. For each item, owner and a yes/no completion check. Add the three steps most often skipped under time pressure.
22. Summarize a long bank or ledger export
Summarize this [bank statement / general ledger] export for [period]: total in, total out, the largest [5] movements each way, recurring payments, and anything unusual for this business type. Flag transactions that look like they need a category review or supporting documentation. Data: [paste redacted export].
Claude handles long, dense exports well; for very large files, summarize in batches and ask it to merge the batch summaries at the end.
4. Audit, Review & Due Diligence
Review prompts tell you what to look at, not what to conclude. The opinion stays with the professional.
23. Generate a risk-based sampling focus
For a [review / audit] of [entity, sector, size], with these characteristics and prior findings: [describe], propose where to focus testing. Rank areas by risk, give the rationale for each, and suggest the kind of procedure that addresses each risk. I will set the actual sample and scope.
24. Draft an analytical review
Here are [N] periods of figures for [entity]: [paste]. Perform an analytical review: compute the standard ratios and period-over-period movements, flag every movement outside [threshold], and for each flag list the questions to put to management. Separate "likely explainable" from "needs evidence."
The techniques you're reading about work. Test your prompts now with Prompt Score and see your score in real time.
I am running financial due diligence on a [target type] for a [buyer]. Deal context: [describe]. Produce a document and information request list grouped by: financial, tax, working capital, debt and off-balance-sheet, contracts, and people. Flag the items that most often hide a deal problem for this kind of target.
26. Summarize findings into a memo
Turn these review findings into a clear memo for [audience: partner / client / buyer]: [paste findings]. Structure: scope, summary of findings ranked by significance, the figures behind each, the implication, and recommended next steps. Neutral, evidence-based tone. Mark anything I flagged as unverified as "to confirm."
27. Find inconsistencies across a document set
Read these documents from the same engagement and identify inconsistencies: contradictions in figures, dates, counterparties, or how a transaction is described. Quote the conflicting passages and note which source is normally more reliable. Documents: [paste redacted set].
28. Draft a management representation point
Based on these findings [paste], draft the specific management representation points I should request for [engagement]. Each point should be precise, tied to a finding, and phrased so a yes/no confirmation is meaningful. Flag any area where a representation is not a substitute for evidence.
5. Client Communication
This is where clarity decides whether the client trusts the number. AI is strong at the first draft in a voice the client will read.
29. Translate a tax development into a client alert
Draft a client alert about [recent development: new rule / rate change / deadline / authority guidance]. Audience: [small business owners / high-net-worth individuals / finance directors]. ~400 words. Structure: what changed, who is affected, what to do and by when, and a one-line offer to discuss. Put a clean disclaimer at the top, not a hedge at the end. I will verify every date and figure before sending.
30. Draft a fee-and-scope email
Draft an email to [client] setting out the fee and scope for [service]. Fee basis: [fixed / hourly / retainer]. Be explicit on what is included, what is extra, and the assumptions. Professional, warm, no jargon. End with a clear next step and what I need from them to start.
31. Explain a tax bill the client did not expect
A client is surprised by a [liability / balance due] of [amount] for [period]. The reasons are: [list]. Draft an email that explains why, in plain language, without sounding defensive. Acknowledge the surprise, give the two or three real drivers, state what (if anything) can still be done, and what we will do differently next year.
32. Draft a deadline-reminder sequence
Write a sequence of three short reminder messages to clients for the [filing / payment] deadline of [date]: one at [4 weeks], one at [1 week], one at [2 days]. Escalate urgency without nagging. Each names exactly what the client must send or do. Keep each under 120 words.
33. Rewrite a tense client email
Rewrite this email to keep the substance but lower the temperature. The reader is [an anxious client / a client who is late with documents / an unhappy client]. Remove anything that reads as blame, keep the necessary firmness on the deadline or fee, and end with one concrete next step. Original: [paste].
34. Prepare talking points for a results meeting
I am presenting [year-end results / a tax position / a planning recommendation] to [client] tomorrow. Key facts: [list]. Give me a one-page talking-point sheet: the headline message, the three things they will care about, the two questions they are likely to ask and how to answer, and the single decision I need from them.
6. Compliance & Regulatory
Compliance prompts map requirements to a client's facts. They give you a working draft, not regulatory certainty.
35. Map a new rule to a client's situation
A new [regulation / tax rule / reporting requirement] has been introduced: [paste summary]. My client is [profile]. Produce a gap analysis: does it apply, what it requires, where the client currently stands, the gap, the deadline, and the effort to close it. Prioritize. I will confirm the rule's current text and dates.
36. Build an AML / KYC onboarding checklist
Produce a client-onboarding due-diligence checklist for [client type] appropriate to [jurisdiction] anti-money-laundering obligations. Group by: identity verification, beneficial ownership, source of funds, risk rating, and ongoing monitoring triggers. Flag the higher-risk indicators I should escalate.
37. Draft an internal compliance procedure
Draft a short internal procedure for [process, e.g. handling a client's late filing / responding to an authority query / approving a planning recommendation]. Include: trigger, steps, who signs off, the record to keep, and the escalation point. Keep it to one page a junior could follow.
38. Analyze a filing requirement against a calendar
Here are the entities and obligations I manage: [paste list with types and periods]. Build a compliance calendar of recurring filing and payment deadlines, grouped by month, flagging the ones with the harshest penalties for lateness. Note where a deadline depends on a rule I should confirm rather than assuming.
39. Draft a response framework to an authority query
A client received this query from [tax authority]: [paste redacted query]. Draft a response framework: what they are really asking, the information needed to answer, the documents to attach, the tone to take, and what NOT to volunteer. I will draft the substantive answer and verify positions; you structure the response.
7. Practice Management
These prompts recover the time you spend on the work you never bill.
40. Prioritize the busy-season backlog
Here is my workload: [paste list of clients, deadlines, and estimated effort]. Propose a priority order based on: hard deadlines, penalty exposure, client value, and effort-to-completion. For each, one line on the ranking. Flag any clash where two deadlines compete for the same week.
41. Draft a capacity plan for the season
I have [N] preparers and this pipeline: [paste]. Sketch a capacity plan: rough allocation of work to people, the weeks that will be the bottleneck, what to pull forward now, and what to consider outsourcing or declining. Flag the single biggest risk to hitting the deadlines.
42. Prepare a new-client intake
I have an intake call with a prospective [client type] tomorrow. Their situation in one line: [describe]. Produce: 15 ordered questions to ask, the red flags to listen for, the documents to request, the conflict and AML checks to run, and a sensible fee approach given the shape of the work.
43. Convert time notes into billing narratives
Convert these rough time notes into clean billing narratives that justify the time without disclosing strategy, each under 25 words and consistent in style: [paste]. Group by client and matter. Flag any entry too vague to bill as written.
44. Draft a process for a recurring task
Document a repeatable process for [recurring task, e.g. monthly bookkeeping for a retail client / quarterly VAT return / annual accounts prep]. Include inputs needed, the ordered steps, the review point, the deliverable, and the typical time. Write it so a new team member can run it on the second read.
45. Summarize the week for a team standup
Here is what moved this week across my clients: [paste notes]. Write a summary for a Monday team standup, organized by what is on track, what is blocked and needs help, and what is at deadline risk. Under 200 words.
8. Professional Development & Knowledge Management
AI is a patient tutor. These prompts make it earn its keep.
46. Brief me on developments in my niche
Brief me on the most significant changes in [area, e.g. VAT on digital services / R&D incentives / a sector's accounting treatment] over the [last quarter]. For each: what changed, who it affects, and one source I should read to go deeper. If something is more noise than substance, say so. I will verify before advising any client.
47. Teach me an adjacent topic well enough to spot the issue
I specialize in [your area]. I need to understand [adjacent topic] well enough to recognize when a client matter touches it and needs a specialist. Teach it in one sitting: the core concept, the sub-points that matter in practice, the hard cases, and the signals in a client's situation that mean "refer this."
48. Turn CPD notes into a usable summary
I took these rough notes at [a CPD session / webinar]: [paste]. Turn them into a clean summary: five key takeaways, the open questions I still have, and a short list of changes to make in my practice as a result.
49. Build a recurring client-review template
Design a template I can run on every client at [annual review / onboarding]: the questions to ask, the data to pull, the planning points to check, the compliance items to confirm, and the deliverable to leave the client with. It should survive 10 clients before I revise it.
50. Draft a positioning piece for the firm
I want to publish a short piece on [topic clients keep asking about]. Audience: [client type]. Draft: a title that is not generic, a 150-word abstract, three takeaways the reader leaves with, and one example I can build out. Confident, useful, not salesy.
How to actually use these
A prompt you run once and discard is worth little. A prompt that becomes part of your weekly workflow compounds. Three moves separate the firms that get marginal value from AI from those that get hundreds of hours back a year:
Save them. Every prompt that returns a usable result goes into a personal prompt library tagged by service line and client type. Next time you tune rather than rewrite.
Score them. When a prompt returns something weak, the problem is almost always the prompt, not the model. Use the six criteria that make a good prompt to rewrite: clarity, specificity, context, structure, examples, role. One iteration usually doubles output quality.
Version them. The first time a prompt survives three different clients is the moment it becomes a template. Pin the version that works and keep the old ones, so you can diff your thinking when the law or the model changes. This is prompts as infrastructure applied to a practice, which is exactly where it belongs.
The firms saving the most time are not using more AI. They are using fewer, better, reusable prompts, organized the way they organize their client files, and they verify every number before it leaves the office.
