Your AI Action Plan

Analyze this data for trends, outliers, and practical next steps. Return: 1. top 5 findings, 2. possible explanations, 3. questions to verify, 4. recommended next actions. Do not assume causes without evidence.

Data:
[PASTE]

In this workbook, analyze [TABLE/RANGE] and answer: What changed most, what looks unusual, which rows need attention, and what chart or pivot would best explain the result?

Use these spreadsheet findings to write a decision memo. Include: headline insight, supporting numbers, likely drivers, caveats, recommended action, and what data should be checked next.

Findings:
[PASTE]

Create a recurring analysis checklist for [DATA TYPE]. Include: required inputs, cleaning checks, metrics to calculate, outlier rules, interpretation questions, caveats, and the final memo format.

Write a weekly [TEAM / CLIENT / PROJECT] report.
Period: [DATE RANGE]
Metrics: [PASTE METRICS]
Highlights: [WHAT WENT WELL]
Risks or blockers: [WHAT NEEDS ATTENTION]
Next actions: [3-5 BULLETS]
Audience: [MANAGER / CLIENT / TEAM]. Keep it factual, concise, and easy to scan.

Analyze this report data before drafting the summary.
Data: [PASTE TABLE OR METRICS]
Return: 1. Top 5 changes, 2. likely explanations, 3. questions to verify, 4. what should be highlighted, 5. what should not be overclaimed.

Turn this exported data into a decision memo.
Audience: [WHO WILL READ IT]
Data: [PASTE CSV OR TABLE]
Return: executive summary, key changes, likely causes, recommended action, risks, and a short appendix explaining assumptions.

Create a reusable monthly review template for [ROLE / TEAM]. It should include: required inputs, KPI table, variance analysis, stakeholder narrative, recommended actions, risks, and a quality checklist before sending.

Summarize this compliance-sensitive document for human review. Return: key facts, deadlines, obligations, missing information, ambiguous language, and questions for the responsible reviewer. Do not give legal, tax, or employment advice.

Document:
[PASTE]

Turn this policy or form instruction into a checklist. Include required fields, approvals, deadlines, evidence needed, common errors, and escalation triggers. Mark anything that needs a qualified human decision.

Text:
[PASTE]

Prepare a reviewer packet for this document. Return: 1. plain-English summary, 2. evidence table with source excerpts, 3. missing information, 4. risk questions, 5. reviewer decision log template. Keep all final decisions blank.

Document:
[PASTE]

Create a reusable compliance-review prompt for [ROLE]. It must require AI to: summarize only, cite source text, flag uncertainty, list missing information, avoid final legal/tax/employment decisions, and produce questions for the qualified reviewer.

Write a client message from these notes.
Client context: [CONTEXT]
Goal: [UPDATE / ASK / EXPLAIN / FOLLOW UP]
Notes: [PASTE NOTES]
Tone: professional, warm, and concise.
Include: clear answer, next step, owner, and deadline if relevant.

Create five client-message templates for my role: status update, request for missing info, delay explanation, decision explanation, and follow-up. Each template should include placeholders, tone guidance, and a checklist for facts I must verify before sending.

Use this client context and draft to prepare a final message.
Context: [PASTE]
Draft: [PASTE]
Check for: factual accuracy, unclear promises, missing deadlines, tone risks, and next step. Return the revised message plus a send/no-send checklist.

Build a client communication style guide for [ROLE / COMPANY]. Include: approved tone, banned phrases, claims that require review, escalation triggers, examples of good messages, and a checklist before sending AI-assisted messages.

Turn these rough process notes into a simple SOP. Include: purpose, when to use it, required inputs, step-by-step workflow, owner, tools, quality checks, and common mistakes.

Notes:
[PASTE]

Create a reusable checklist for [PROCESS]. Include sections for intake, execution, quality check, handoff, escalation, and completion evidence. Format it so I can paste it into Notion.

Convert this process explanation into: 1. SOP, 2. quick checklist, 3. training note for a new person, 4. edge cases, 5. questions to confirm with the process owner.

Transcript or notes:
[PASTE]

Design a procedure-library template for [TEAM]. Include fields for owner, last reviewed, inputs, steps, tools, examples, risks, escalation path, related templates, and update log. Add rules for when a procedure must be reviewed.

N/A — manual close checklist.

You are helping an accountant run a materiality-focused month-end close review. Do NOT recommend any posting — flag only.

Inputs:
- Prior-period trial balance: [PASTE]
- Current-period trial balance: [PASTE]
- Materiality threshold (absolute $): [AMOUNT]
- Materiality threshold (% change): [PERCENT]
- Known one-off events this period: [LIST OR 'none']

Return:
1) Accounts over the $ threshold with amount and % change.
2) Accounts over the % threshold with amount and absolute change.
3) For each flagged account, a one-line plausible explanation and a specific question the accountant should ask the client.
4) Items the AI cannot explain from the data — label 'needs human review'.

You are drafting a month-end close memo for a controller's review. Keep it concise — bullet form only — and clearly separate facts from questions.

Inputs:
- Flux summary exported from [Xero / QuickBooks Advanced / Sage Intacct]: [PASTE]
- Uncategorised transactions list: [PASTE]
- Client: [CLIENT NAME / INDUSTRY]
- Known context this period: [NOTES]

Return:
1) Flux highlights (top 10 movements with one-line explanation).
2) Uncategorised transactions remaining (with suggested GL code and confidence).
3) Open questions for the controller.
4) Items requiring CPA review before the books are closed.

You are scoring one account's flux for a multi-entity month-end close. Return JSON only.

Inputs:
- account_code, account_name
- prior_balance, current_balance
- period, entity_id
- materiality_rules: { absolute_usd, percent_change }
- known_events: [...]

Return:
{
  "flag": "material|immaterial|review",
  "delta_usd": 0,
  "delta_pct": 0,
  "explanation_candidates": ["..."],
  "reviewer_questions": ["..."],
  "confidence": "low|medium|high"
}

Do not output booking recommendations. If inputs look inconsistent (e.g. opening balance mismatch), set flag='review' and return a blocking note.

N/A — handled in tax software.

You are helping a CPA prepare for a client tax return. Do NOT compute tax liability — organize inputs only.

Inputs:
- Prior-year return highlights: [PASTE summary]
- Client-submitted documents this year: [LIST]
- Known life/business changes: [LIST OR 'none reported']
- Jurisdictions: federal + [STATE(S)]

Return:
1) Documents that appear to be missing based on last year's return.
2) New deductions or credits the client may qualify for, each with the IRS form or publication reference.
3) Client questions needed before preparer can start.
4) Items flagged as 'CPA must verify' (complex or judgment calls).

Draft a client-facing tax return review memo in plain language. The return is already prepared — you are helping the CPA explain it.

Inputs:
- Return summary (AGI, total tax, refund/balance due): [PASTE]
- Notable changes vs. prior year: [LIST]
- AI assistant diagnostic flags: [PASTE]
- Planning opportunities identified: [LIST]
- Client name and filing status: [DETAILS]

Return a memo with:
1) Bottom line (refund/balance due) in one sentence.
2) Year-over-year delta drivers.
3) Items resolved and items still open.
4) Proactive planning suggestions for next year.
5) Confirmation items the client must sign off before e-file.

You are reviewing one prepared tax return for potential errors and missed opportunities. Return JSON only.

Inputs:
- return_type (1040, 1120S, 1065, etc.)
- jurisdictions: [federal, state(s)]
- return_summary (key totals)
- schedule_summary (by schedule)
- prior_year_return_summary
- firm_review_rules: [list of firm-specific flags]

Return:
{
  "preparer_errors": [{"issue": "...", "line_reference": "...", "severity": "low|medium|high"}],
  "missed_credits_or_deductions": [{"item": "...", "irc_or_state_ref": "...", "confidence": "low|medium|high"}],
  "jurisdiction_edge_cases": [{"state": "...", "flag": "..."}],
  "reviewer_questions": ["..."],
  "overall_confidence": "low|medium|high"
}

Do not output 'approved' — only findings. Any item with confidence < medium must be labelled 'needs CPA verification'.

N/A — manual Excel mapping.

You are drafting first-pass financial-statement commentary for a CPA to review. Do NOT assert compliance — draft only.

Inputs:
- Current-period statements (BS, P&L, CF): [PASTE]
- Prior-period statements: [PASTE]
- Entity type: [Private / Public / Non-profit]
- Reporting framework: [US GAAP / IFRS / other]
- Known events this period: [LIST]

Return:
1) MD&A style commentary (revenue drivers, margin, cash story, significant items) — 200-300 words.
2) Footnote skeleton: topic headers and one-line scope for each (related parties, commitments, contingencies, subsequent events, etc.).
3) Disclosures likely required under [framework] for this entity type.
4) Items the CPA must confirm before finalizing.

Draft client-facing financial-statement commentary in the firm's house style. Inputs below.

Inputs:
- Reporting pack from [Sage Intacct / NetSuite / QuickBooks Advanced]: [PASTE]
- House style guide (tone, voice, sentence length): [PASTE OR LINK]
- Client context: [SHORT PROFILE]
- Prior-period commentary (for consistency): [PASTE]

Return:
1) Cover letter (2-3 paragraphs).
2) MD&A commentary aligned to the reporting pack.
3) Footnote draft language where the pack already contains data.
4) A reviewer checklist — what the CPA should confirm before releasing.

You are checking one financial-statement pack against a firm disclosure matrix. Return JSON only.

Inputs:
- entity_type (Private / Public / Non-profit / Employee-benefit)
- reporting_framework (US GAAP / IFRS / special-purpose)
- jurisdictions
- statement_pack: balance_sheet, income_statement, cash_flow, equity_roll_fwd, footnotes_included
- disclosure_matrix: [{topic, required_if, framework_ref}]

Return:
{
  "missing_disclosures": [{"topic": "...", "framework_ref": "...", "reason": "..."}],
  "likely_extra_disclosures": [{"topic": "...", "framework_ref": "...", "confidence": "low|medium|high"}],
  "formatting_issues": ["..."],
  "reviewer_questions": ["..."]
}

Do not output 'ready to issue' — only findings. Anything novel or framework-specific must be flagged as 'partner review required'.

N/A — manual pivot work.

You are drafting a variance narrative for a monthly business review. Do NOT invent explanations — if the variance is unexplained, flag it as 'owner must explain'.

Inputs:
- Actuals-vs-budget table (account, department, actual, budget, variance_$, variance_%): [PASTE]
- Materiality threshold ($ and %): [VALUES]
- Prior-period variance notes: [PASTE]
- Known one-offs: [LIST OR 'none']

Return:
1) Top 10 material variances ranked by absolute $.
2) For each, a proposed plain-English explanation grounded in data provided (or 'needs owner explanation').
3) Questions the controller should ask each department owner.
4) A 3-sentence exec summary at the top.

Polish this raw variance analysis into a management-review narrative. Keep facts intact; improve clarity and tone.

Inputs:
- Raw Copilot variance output: [PASTE]
- Prior-period narrative for consistency: [PASTE]
- Target audience: [Exec team / Board / Dept heads]

Return:
1) Exec summary (3 sentences).
2) Top variances with cause, action, owner.
3) Forward-looking risks and opportunities.
4) Questions for next month's prep.

Generate a structured variance bot message for Slack. Return markdown only.

Inputs:
- Entity, period, actuals-vs-budget table (JSON)
- Department owners map (department → Slack handle)
- Firm materiality rules
- Prior-period flagged items

Return:
- One-liner headline (net variance and direction).
- Top 5 material lines with cause and owner tag.
- Open questions list.
- Data freshness and source statement.

Do not fabricate explanations. If the bot cannot find a documented cause in provided notes, mark 'owner explanation required'.

N/A — manual sampling.

You are helping an audit associate draft a workpaper exception narrative. Do NOT conclude audit opinions.

Inputs:
- Test name and population: [DETAILS]
- Sample selection method: [DETAILS]
- Raw exception notes: [PASTE]
- Materiality: [AMOUNT]
- Control being tested: [DETAILS]

Return:
1) Exception summary grouped by exception type.
2) Projected misstatement computation (show formula).
3) Risk implications for the control objective.
4) Open items for the audit senior.
5) Items that must be referred to the engagement partner.

Draft workpaper narratives for flagged anomalies. Keep wording conservative and clearly separate facts from follow-up recommendations.

Inputs:
- Anomaly list from [MindBridge / DataSnipper]: [PASTE]
- Engagement context: [CLIENT TYPE / INDUSTRY / SIZE]
- Applicable framework: [GAAS / PCAOB / ISA]
- Prior-year equivalent narrative (style anchor): [PASTE]

Return:
1) Per-anomaly narrative (fact, relevance, proposed follow-up).
2) Aggregated pattern summary (if any).
3) Recommended procedures for the senior.
4) Items to escalate to the partner.

You are reviewing one audit workpaper for completeness and internal consistency. Return JSON only.

Inputs:
- workpaper_id, program_step_ref
- workpaper_text
- related_workpaper_refs
- firm_review_rules

Return:
{
  "program_step_coverage": {"covered": true|false, "gaps": ["..."]},
  "tie_reference_issues": ["..."],
  "internal_inconsistencies": ["..."],
  "language_flags": ["uses 'verified' without evidence", "opinion-style wording"],
  "senior_review_items": ["..."]
}

Do not output 'workpaper approved' — only findings. Any item involving conclusions or opinions must be flagged 'partner review required'.