These Are Not "Write Me an Email" Prompts
The internet is full of AI prompt lists that are about as useful as a screen door on a submarine. "Write me a professional email." "Summarize this article." "Give me 10 ideas for social media posts." Congratulations, you have replicated what anyone with a ChatGPT or Claude account figured out in their first five minutes.
The prompts below are different. These are the actual prompts we use at OneWave AI for real client work. They are specific, structured, and designed to produce outputs you can actually use without spending 30 minutes editing. Copy them, paste them, customize the bracketed sections, and watch the difference between a generic prompt and a well-engineered one.
For each prompt, we explain why it works and how to customize it for your specific situation.
1. Summarizing a Meeting Transcript Into Action Items
I am going to paste a meeting transcript below. Please analyze it and produce the following: (1) A one-paragraph executive summary of the meeting's purpose and outcome. (2) A numbered list of every action item mentioned, with the responsible person's name and any stated or implied deadline. (3) A list of decisions that were made during the meeting. (4) A list of open questions or unresolved items that need follow-up. Format each section with a clear heading. If a responsible person or deadline is unclear, flag it explicitly. [Paste transcript here]
Why it works: The numbered output structure forces the AI to categorize information rather than just summarize it. Asking it to flag unclear responsibilities catches the ambiguity that plagues most meetings. You get something you can paste directly into a project management tool.
Customization tip: Add your project management format at the end -- "Format action items as Asana tasks with assignee, due date, and project name" -- and the output drops straight into your workflow.
2. Analyzing a Competitor's Website and Identifying Gaps
I want you to act as a competitive analyst. Here is the URL and a description of our competitor's website: [competitor details]. Here is a description of our company and what we offer: [your company details]. Please analyze the competitor and produce: (1) Their core value proposition in one sentence. (2) Their target audience based on messaging and content. (3) Strengths -- what they do well in terms of positioning, messaging, and apparent offerings. (4) Weaknesses -- gaps in their messaging, missing content, unclear positioning, or underserved segments. (5) Opportunities for us -- specific ways we could differentiate or capture attention they are missing. Be specific and cite examples from their site where possible.
Why it works: Instead of asking for a vague "analysis," this prompt structures the output around actionable intelligence. The five categories mirror what a strategy consultant would deliver, and asking for specific examples prevents the AI from generating generic observations.
Customization tip: Paste the actual text from the competitor's homepage, about page, and pricing page rather than just the URL. The more raw material you give the AI, the sharper the analysis.
3. Turning Raw Customer Feedback Into a Prioritized Feature List
Below is a collection of raw customer feedback from [source: support tickets, surveys, reviews, etc.]. Please analyze this feedback and produce: (1) A categorized list of every distinct feature request, bug report, and complaint, grouped by theme. (2) A frequency count of how many times each theme appears. (3) A prioritized list ranking themes by frequency and apparent business impact, with a brief justification for each ranking. (4) Three "quick wins" -- items that appear frequently and seem straightforward to address. (5) Three "strategic bets" -- items that are mentioned less often but could be significant differentiators. Be specific. Quote directly from the feedback where relevant. [Paste feedback here]
Why it works: Raw feedback is noisy. This prompt forces the AI to do the categorization, counting, and prioritization that a product manager would spend hours on. The "quick wins" and "strategic bets" distinction is especially useful because it separates urgent fixes from long-term opportunities.
Customization tip: Add context about your current product roadmap so the AI can flag overlap -- "Here is our current roadmap for reference: [roadmap]. Note which feedback items are already planned."
4. Writing a Follow-Up Email After a Sales Call
I just finished a sales call with [prospect name] at [company]. Here are my notes from the call: [paste notes]. Please write a follow-up email that: (1) Thanks them for their time and references one specific topic we discussed to show I was listening. (2) Summarizes the key points we agreed on. (3) Addresses the main objection or concern they raised, with a brief, confident response. (4) Proposes clear next steps with a specific date or timeframe. (5) Keeps the tone professional but warm -- not salesy, not robotic. Keep it under 200 words. End with a single clear call to action.
Why it works: The specificity prevents the AI from generating a generic "great chatting with you" email. By feeding in your actual call notes and asking it to reference specific discussion points, the output feels personal. The word limit prevents the AI from writing a novel nobody will read.
Customization tip: Add your company's email signature format and any specific collateral you want to reference -- "Include a link to our case study on [topic] since they mentioned struggling with that." For more on using AI throughout the sales process, see our guide on how to use Claude for sales.
5. Creating an SOP From a Verbal Process Description
I am going to describe a process that our team follows verbally. Please turn it into a formal Standard Operating Procedure (SOP) document with the following structure: Title, Purpose (one sentence on why this process exists), Scope (who this applies to), Prerequisites (what needs to be in place before starting), Step-by-Step Procedure (numbered, with enough detail that someone new could follow it), Common Mistakes to Avoid, and Escalation (when and how to escalate if something goes wrong). Use clear, direct language. If my description is ambiguous or seems to have gaps, flag them with [CLARIFICATION NEEDED] and explain what information is missing. Here is the process: [describe the process]
Why it works: Most SOPs never get written because the person who knows the process does not have time to document it. This is exactly the kind of institutional knowledge that belongs in an AI knowledge base. This prompt lets them brain-dump verbally (or in rough notes) and produces a structured document. The "flag ambiguity" instruction is critical -- it catches the gaps that the person doing the brain-dump does not realize they are skipping.
Customization tip: If your company has an SOP template, paste it into the prompt and say "Follow this exact format" instead of using the structure above.
6. Reviewing a Contract for Red Flags
Please review the following contract and identify potential issues. For each issue found, provide: (1) The specific clause or section number. (2) What the clause says in plain English. (3) Why it could be problematic. (4) A suggested revision or negotiation point. Focus especially on: auto-renewal terms, liability limitations, indemnification clauses, termination conditions, intellectual property ownership, non-compete or exclusivity language, payment terms and penalties, and any language that is vague or one-sided. This is not legal advice -- I will review your findings with my attorney. I need this as a starting point for discussion. [Paste contract here]
Why it works: The disclaimer keeps expectations appropriate -- this is a first-pass analysis, not legal counsel. Listing the specific areas to focus on prevents the AI from giving you a generic summary and forces it to look at the clauses that actually matter in business contracts. The "plain English" requirement ensures you can understand the findings without a law degree.
Customization tip: Add industry-specific concerns -- "This is a SaaS vendor contract, so also check for data ownership clauses, SLA terms, and what happens to our data if we terminate."
7. Generating a Weekly Status Report From Project Notes
Here are my raw project notes from this week: [paste notes]. Please generate a weekly status report with the following sections: (1) Summary -- two to three sentences on overall progress this week. (2) Completed This Week -- bullet list of what got done. (3) In Progress -- bullet list of what is actively being worked on, with estimated completion. (4) Blocked or At Risk -- anything that is stuck or could cause delays, with the reason why. (5) Next Week Priorities -- the top three to five items for next week. (6) Key Metrics or Numbers -- any relevant data points from my notes. Keep it concise. Use the tone of a senior manager reporting to leadership -- factual, no fluff, focused on outcomes.
Why it works: Status reports are universally hated and universally necessary. This prompt transforms messy weekly notes into a polished report in the format that leadership actually wants to read. The "blocked or at risk" section is especially valuable because it forces visibility on problems before they become crises.
Customization tip: Add the names of your projects or workstreams at the beginning so the AI can organize notes by project rather than chronologically.
8. Building a Job Description From a Rough List of Responsibilities
I need to hire for a role and here is what I know about it: [paste rough notes about the role, responsibilities, and requirements]. Please create a complete job description with: (1) A compelling job title (suggest two to three options). (2) A two to three sentence overview that sells the role without being cringeworthy. (3) Key Responsibilities -- bullet list, starting each with an action verb. (4) Required Qualifications -- what someone must have, being realistic not aspirational. (5) Preferred Qualifications -- nice-to-haves, clearly labeled as such. (6) What success looks like in the first 90 days. (7) Salary range context -- if I provide a range, include it. If not, suggest where to research market rates. Avoid corporate buzzwords. Do not use "rockstar," "ninja," "fast-paced environment," or "wear many hats." Write like a human.
Why it works: The explicit ban on corporate cliches produces job descriptions that actually attract good candidates instead of making them cringe-scroll past your listing. The "90-day success" section is a powerful addition that most job descriptions lack -- it tells candidates what winning looks like and helps you clarify your own expectations.
Customization tip: Add a description of your company culture and team dynamics so the AI can weave that context naturally into the description.
9. Analyzing a P&L Statement and Flagging Anomalies
I am pasting our P&L statement below. Please analyze it and provide: (1) A plain English summary of our financial performance this period. (2) Key ratios -- gross margin, operating margin, net margin, and how they compare to typical benchmarks for [your industry]. (3) Anomalies -- any line items that look unusual, have changed significantly from the prior period, or seem out of proportion. For each anomaly, explain what it might indicate and what questions I should ask. (4) Three areas where we could potentially reduce costs or improve margins, based on what you see. (5) One-paragraph outlook based on the trends in the data. This is for internal discussion, not audit purposes. Flag anything that warrants a deeper look with our accountant. [Paste P&L here]
Why it works: Most business owners look at P&L statements and see numbers. This prompt turns those numbers into a narrative with specific questions to investigate. The industry benchmarking adds context that makes the numbers meaningful. Flagging items for your accountant keeps the analysis in its proper lane.
Customization tip: Include prior period P&L data for comparison -- "Here is this quarter and last quarter. Highlight significant changes between the two."
10. Creating a Client Onboarding Checklist From Past Onboarding Notes
Here are my notes from the last several client onboardings we have done: [paste notes from multiple onboardings]. Please analyze these and create: (1) A master onboarding checklist, organized chronologically from pre-kickoff through first 30 days. (2) For each checklist item, note who is responsible (us vs the client) and estimated time required. (3) A list of common issues or delays we encountered and how to prevent them. (4) A "Day One Email" template we can send to new clients that sets expectations and outlines next steps. (5) Success criteria -- how we know onboarding is complete and the client is set up for success. Identify any steps that appeared in some onboardings but not others, and recommend whether they should be standardized.
Why it works: Onboarding processes tend to live in people's heads and evolve organically. This prompt mines your historical experience and turns it into a repeatable system. The "common issues" section is gold -- it turns past mistakes into preventative measures. The standardization recommendation helps you decide which ad-hoc steps should become permanent.
Customization tip: Include feedback from clients about their onboarding experience to capture the client's perspective, not just your internal view.
The Pattern Behind Every Good Prompt
If you study the prompts above, you will notice a pattern. Every one of them does four things:
- Sets the context -- tells the AI what role it is playing and what information it is working with.
- Specifies the output structure -- defines exactly what sections and format you want, so you do not get a wall of text.
- Defines quality standards -- includes phrases like "be specific," "cite examples," "flag ambiguity," or "keep it under 200 words."
- Sets boundaries -- clarifies what the output is for and what it is not (e.g., "not legal advice," "for internal discussion").
Apply that framework to any business task and you will write prompts that produce genuinely useful output. If you want a deeper look at the tool we use behind most of these prompts, read our practical guide to Claude AI for business. The difference between a mediocre prompt and a great one is not magic -- it is structure.