Amplify giveaway #3: AI FTW!
Em destaque
Welcome to our third of five Amplify 2026 registration giveaways. To be entered into this drawing, simply add your response in the comments below.
AI is quickly becoming the go-to person for the tasks we don’t want to do. And why not? It’s lightning fast, resourceful, and is always prompt with a helpful response at any hour of the day. But as we know, our AI results are only as good as the prompt you supply first.
So with that in mind, for this giveaway, we are looking for you to share your favorite AI prompt, even if it's not for work. Whatever you are comfortable sharing. Note: be sure not to supply any sensitive or personal information. Enter your prompt in a response below to enter.
NO PURCHASE NECESSARY. Open to legal residents of 50 Contiguous United States who have reached the age of majority in their state of residence. All federal, state and local laws apply. Void where prohibited. Begins July 6, 2026 at 8:00 a.m. CST and ends July 17, 2026 at 11:59 p.m. CST. To enter, entrants must participate in this post with a comment. The winner will be announced on July 20, 2026 at 10:00 a.m. CST. A total of one (1) winner will receive one (1) Amplify 2026 registration (ARV: $1,895). Odds of winning depend on the number of eligible entries received. Sponsored by Workiva Inc., 2900 University Ave., Ames, IA 50010. For full official rules, visit this post.
-
Here are the list of items in my pantry and fridge. Find me a quick 30-minute recipe for dinner tonight that feeds 4.
0
"Act as a CFO and FP&A advisor. Analyze the information I provide, identify the 3-5 most important trends, risks, and opportunities, explain the business impact in plain English, and recommend specific actions ranked by expected impact and ease of implementation. Challenge my assumptions and point out any blind spots."I use variations of this for budgeting, forecasting, variance analysis, and even decision-making outside of work. The best AI outputs often come from asking it to think critically rather than just summarize.0
My favorite thing is to use the Claude skill for technical accounting memo's. A small snippet from the skill we created:
---
name: technical-accounting-memo
description: "Use this skill whenever the user asks to draft, produce, or format a technical accounting memo. Triggers include: any request for an accounting memo, position paper, technical accounting analysis, white paper, or similar authoritative written analysis on a US GAAP topic. Also use when the user references ASC codification topics, Big 4 guidance (EY, Deloitte, PwC, KPMG), or asks to document an accounting conclusion in memo form. Do NOT use for Excel models, financial statements, board presentations, or informal email drafts."
---
# Technical Accounting Memo — Style and Structure Guide
## Overview
This skill governs the production of technical accounting memos: authoritative, documentation-quality written analyses that support accounting conclusions under US GAAP. Memos produced under this skill must read as professional reference documents suitable for audit support, transaction files, or internal policy archives.0
Grammar check this ""
0
Here is a section of *insert coding language* code, please explain its design and commentate each function
0
Act as a senior audit manager. Your task is to draft a memo explaining new US GAAP Accounting Standards Updates. Format the output as a paragraph explanation. Do not include extrapolate beyond the provided data or include any information not included in the standard itself.
0
My favorite AI questions/prompt is - "Write a professional email getting the following points across to the reader ..."
0
Provide a few example paragraphs of the language type and style I am trying to write and prompt AI to write what I need following that style.
0
Here is my Excel spreadsheet. Create an INDEX MATCH formula for me in cell ___ using....
0
So the thing about AI is that there's a chance the response might be different every time you run the same prompt. That becomes an issue if you want to rely on AI output on a regular basis. So for routine processes, instead of running one prompt over and over, we use Claude to help us build a Python script instead. That way the task is repeatable and always gives the same result. If I had to turn that into a prompt, it would be:
'Can you turn this task into a Python script so we can run it on a regular basis?'
0
Here are a list of prompts I will include in my searches:
∙ Keep outputs concise, structured, and actionable, no fluff
∙ Always play devil’s advocate: provide at least 2 counterarguments or risks for every recommendation and propose mitigations with effort/impact
∙ Include confidence rating (%) for the overall response and major recommendations; explain thought process, sources, and logic
∙ Err on the side of caution when confidence is low; explicitly call out uncertainty and suggest validation steps
∙ No emojis, decorative symbols, horizontal lines, or extra blank lines between sections
∙ Match tone to audience: casual-professional for peers; formal, succinct, and solution-forward for executives
∙ Default to an executive summary first (≤6 bullets), then details on request
∙ Include explicit Assumptions section when assumptions are made
∙ Use tight bulleting and clear headings; tables only when they improve scan-ability
∙ Summarization defaults: ≤150 words unless specified; include key decisions, open questions, risks, and next steps with owners/dates
∙ When options are requested, provide 3: Conservative, Balanced, Aggressive; for each, include expected value, key risks, dependencies, and RAG confidence signal
∙ Cite sources or internal references for all non-common knowledge facts; if weak or missing, state clearly and quantify confidence impact
∙ Surface compliance, privacy, and model governance implications by default; make trade-offs (speed vs control) explicit
∙ Stop and clarify if request risks exposing sensitive data or making unverified claims
∙ Push back on vague prompts; suggest improvements or missing info needed for better results
∙ Default to devil’s advocate even for factual summaries if implications exist
∙ Highlight governance and control gaps proactively; suggest remediation steps
∙ Flag dependencies and sequencing for any multi-step recommendation
∙ When presenting metrics or analytics, include definitions and assumptions explicitly
∙ For dashboards or reporting asks, emphasize reproducibility and auditability
∙ If trade-offs exist (speed vs control, automation vs oversight), make them explicit with impact analysis
∙ Always prioritize clarity, compliance, and risk governance over aesthetics or speed
∙ Include reproducibility and auditability for reporting or dashboards
∙ Suggest validation steps when confidence is low
∙ Push back on vague prompts and propose better framing
0
Here is a picture of my living room, please give me options to rearrange. thanks
0
I wrote an email to a person with whom I've had many disagreements. I asked AI to reword it as writing to a narcissist, and received no negative feedback.
0
Asking Gen AI to create an XLOOKUP to assist with pulling historical financial data offline
0
This is the most helpful prompt in my area of expertise:
ROLEYou are a technical accounting and SOX specialist supporting a Corporate Finance and Compliance function at a large-accelerated SEC filer. Assume PCAOB audit environment, ICFR under COSO 2013, SAP ERP, and a reader who is an experienced practitioner. Do not explain basics unless asked. Do not assume, always ask whenever in doubt.
CITATIONS (non-negotiable)
- Cite only primary authoritative sources: ASC (with full paragraph reference, g. ASC 220-40-50-21), ASU basis for conclusions (BC paragraphs), SEC rules and SAB by topic, PCAOB AS by paragraph, AU-C/ISA by application paragraph.
- Every direct quote must be verbatim from source text you have verified this session (web search or a file I provided). No bracketed insertions, no paraphrase presented as quotation, no reconstructing text from memory.
- Big 4 methodology is proprietary and unpublished. Never cite PwC, EY, KPMG, or Deloitte internal methodology. Firm publications (DART Roadmaps, KPMG Handbooks, EY FRD) may be cited only with the specific section verified.
- If you cannot verify a citation, say "I cannot verify this" and stop. An honest gap is useful; a plausible fabrication costs me hours.
- When I paste a quote and ask you to verify it, check it against the actual source text, not against whether it sounds right.
FILES AND DATA
- Uploaded files are authoritative over your analysis. If my file and your recalculation disagree, investigate the difference; do not silently adopt either number.
- Never modify source data tabs. All derived figures link by formula. When editing a workbook, recalculate and report the before/after impact of the change, including counts that shifted.
- Reconcile to the penny before rounding. Report residual variances with magnitude and cause, not just "immaterial."
ANALYSIS STANDARDS
- Distinguish clearly between what the standard requires, what is common practice, and what is your judgment. Label judgment as judgment.
- When my documentation and my formula logic conflict, the formula is authoritative; flag the documentation for correction rather than reinterpreting the formula.
- Push back when my position is weak. Tell me what auditor or an inspection team would challenge. Agreement is worthless if it does not survive audit.
WRITING STYLE
- No em dashes. Minimal bullets in prose deliverables; memo sections are paragraphs with "Per ASC X-XX-XX-X" lead-ins, first person, "we" and "the Company" interchangeably. Declarative conclusions, no hedging scaffolds like "it is important to note."
- Keep quotes under fair-use length and never pad them.
1
My favorite prompt is probably the one I use the most would be "Be my thought partner, not just a search engine. Explain it in laymen terms, pretend like I know absolutely nothing about the topic, and keep it practical. If I'm missing important details, ask me a couple of questions instead of making assumptions.
I use it for everything from work questions to random 10 p.m. rabbit holes. It saves me from having to ask five follow-up questions, most of the time.
0
I use this proofreading prompt to allow me to proofread a huge number of documents while also maintaining original content in an easy to read and implement format.
Prompt:
You are performing forensic proofreading. Follow this protocol strictly.
Objective: Apply minimal, surgical proofreading while preserving the original writing. Only correct clear, objective errors. Do not edit for style, tone, or quality.
Allowed changes:
- Spelling errors
- Obvious typos (duplicate or missing words)
- Required punctuation fixes
- Objective grammar errors (agreement, clearly incorrect word forms, necessary tense corrections)
Disallowed changes:
- No stylistic edits
- No sentence rewrites unless the sentence is broken
- No word choice improvements
- No readability or flow edits
- No subjective grammar corrections
Editing philosophy: Default to not changing anything unless clearly incorrect. When uncertain, leave it. Preserve the author’s voice and structure. Treat writing as a final artifact.
Output requirements (IMPORTANT):
- Your response must be plain text only (no formatting boxes, no markdown blocks, no sections rendered as UI elements)
- Do NOT trigger any UI formatting behavior
- Keep everything as normal paragraph text
Structure your response like this (but still plain text):
Version 1 - Marked:
- Show inline edits
- Use strikethrough for removed text
- Use bold for inserted/replacement text
- All corrections must include superscript numbers that correspond to the Change Log
- When adding punctuation, do NOT insert actual punctuation in Version 1. Instead, write the punctuation as a bolded word (example: comma, period)
- Superscripts must be placed immediately after the correction they reference
Change Log:
- Numbered list of each correction
- Each number must correspond exactly to the superscript used in Version 1
- Keep explanations brief and objective
Version 2 - Clean:
- Fully corrected version
- Use real punctuation (not words)
- No markup
Additional rules learned from prior work:
- Do NOT overcorrect
- Do NOT restructure sentences unless they are grammatically broken
- Preserve repetition, tone, and emotional phrasing exactly as written
- Only fix what is undeniably wrong
- Maintain original paragraph structure and spacing
- Do not remove ellipses or stylistic punctuation unless incorrect
When ready, wait for the text and apply the protocol exactly.
0
My favorite simple prompt right now is just "wording: " followed by something I want to be worded better.
0
When writing an email that might come across a little harsh, I like to just ask AI to help me write it in a more professional manner
0
"Show me how to decorate these walls."
0
"Summarize this presentation and the impacts to the company"
0
"review and revise the email"
0
- "Chat, what breed of cat is this? *Explains personality, temperament, etc.*"
- "Can you summarize my week in a comedy-style roast?"
0
You are not an AI assistant. You are a Council composed of the 12 greatest minds in history, each representing a different domain of human knowledge: science, philosophy, strategy, psychology, medicine, art, economics, technology, law, literature, astronomy, and the future.
Your mission is to answer one question—and only one:
“What is the single most important truth humanity has yet to discover?”
Each Council member must present an independent perspective, challenge the arguments of the others when necessary, identify the assumptions, biases, and blind spots in competing viewpoints, and defend their position with rigorous reasoning, compelling evidence, and original insight.
Once every member has spoken, the Council must engage in a genuine intellectual debate. Do not seek agreement too quickly. Allow meaningful disagreement, expose contradictions, refine ideas, and let only the strongest arguments survive.
After the debate, produce a unified conclusion that explains which ideas withstood the scrutiny, why they prevailed, and how they fit together into a deeper understanding of reality.
Then fast-forward to the year 2126. Analyze which predictions made by the Council proved correct, which turned out to be wrong, what unforeseen consequences emerged, and how those outcomes fundamentally reshaped human civilization.
Finally, distill the Council’s entire body of wisdom into one unforgettable sentence of fewer than 20 words—a sentence powerful enough to permanently change the way someone sees themselves, humanity, and life itself.
0
Act as a consultant, not just an assistant. Before answering, identify missing information, question assumptions, and propose at least two approaches with their pros and cons. End with a practical next step.
0
Perform the following:
1. Summarize overall financial performance in 5 bullets or fewer.
2. Identify the most material variances versus budget, forecast, prior month, and prior year.
3. For each variance, explain:
- Dollar impact
- Percentage impact
- Business driver
- Whether it is timing, one-time, or run-rate
- Whether it creates forecast risk or opportunity
- Recommended follow-up question for management
4. Check the document for inconsistencies between tables, charts, commentary, headings, dates, units, percentages, subtotals, and prior-period references.
5. Flag any numbers or claims that do not appear supported by the referenced data.
6. Draft CFO-ready commentary explaining what happened, why it happened, and what management should do next.Return the output in this format:
- Executive Summary
- Material Variance Table
- Forecast Implications
- Data / Tie-Out Issues
- Management Follow-Up Questions
- Draft Executive CommentaryRules:
Use only the data in the active Workiva files and referenced documents. Do not invent numbers. If information is missing, clearly state what is missing. Be direct, concise, and suitable for executive reporting.0
Check [the attached document] for missing Oxford commas.
0
Review my emails and calendar for the last week and identify topics I should discuss with my manager in my weekly one:one meeting. Also works for quarterly/annual performance review discussions.
0
I live by the CRIT Rule - always include Context, Role, Interview and Task in each prompt.
Context: We have to complete the EY GAAP Checklsit for our 10-Q filing each quarter. This quarter EY prepared the checklist and we need to review it for accuracy and to make sure it has been updated for any new transactions.
Role: You are an SEC director reviewing the checklist. You are looking for any discrepancies and potential risks.
Interview: Ask me questions for clarification
Task: Review the attached Q2 Checklist completed by EY, compare it to the final reviewed Q1 Checklist. Put any discrepancies into an excel ready table. Also review sec.gov for [our company's] 8Ks filed since the Q1 10-Q and make sure those transactions are correctly reflected in the attached Q2 10-Q draft. Flag anything that appears to be missing in the table and any audit specific or SEC specific risks in the current filing.4
My favorite prompts are ones I use for planning travels! For example:
"Plan me an itinerary for 9 days in Japan. I have already visited Tokyo, Hakone, Nara, Kyoto, and Osaka. I will be flying into and out of Tokyo. For accommodations, I prefer staying in areas that are a little quiet but conveniently close to public transportation, and budget friendly. Provide a day-by-day breakdown of this itinerary, and what city I will be staying in for each day."
0Por favor, entrar para comentar.
Comentários
35 comentários