Finance work doesn’t have to live in Excel. I drafted these three prompts for the boring but important stuff — variance reports, expense audits, cash flow forecasting. Copy them, drop in your data, tune from there.
Quick caveat before you start: these are accelerators, not replacements. AI writes a passable first draft of a finance report but it misses subtle stuff. Always have a human sanity-check the numbers before anything leaves the department.
I put these together from vendor docs, G2 and Capterra reviews, plus a few Reddit threads where actual controllers talk about what works in practice. I haven’t run them in a live accounting system, so treat the prompts as starting points and pressure-test on your own data.
[Claude]
Variance report generator. Feed it a CSV of actuals vs budget by department and it’ll spit out a markdown summary table, the top over-budget items, and three questions per department head. Anything over 10% variance gets bolded.
You are a senior FP&A analyst. Given a CSV of monthly actuals vs budget by department (columns: month, department, account, actual_usd, budget_usd), produce a markdown variance report with: (1) a summary table of total actual, total budget, and variance % per department, (2) the top 5 over-budget line items with dollar amount and likely cause, (3) three questions to ask each department head. Flag any variance over 10% in bold.
[ChatGPT]
Expense audit prompt. Returns JSON for each flagged violation with severity rating and a one-line action. Run it as a first pass — a human reviews before anything escalates.
Act as a finance controller auditing employee expense reports. Given a list of expenses (date, employee, amount_usd, category, receipt_text), return a JSON array where each entry flags items that violate this policy: meals over $75/person, hotels over $250/night, any missing receipt over $25, and any weekend entertainment spend. Include for each flag: rule_broken, severity (low/med/high), and a one-sentence recommended action.
[Cursor]
Cash flow projection in Python. Twelve months of transactions in, 90-day forecast out. Pandas and stdlib only — no extra dependencies, type hints and a usage example baked in.
Write a Python function forecast_cash_flow(transactions_csv: str, days_ahead: int = 90) -> dict that loads 12 months of transactions, computes a 30-day moving average of net daily cash flow, and projects the next 90 days. Return a dict with keys: projected_balance, lowest_day, lowest_balance, runway_days. Use only pandas and stdlib. Include type hints, docstring, and a 3-line usage example at the bottom.
Full archive at aitoolalliance.com/tag/prompt-pack.