What is in this pack
This page covers chatgpt prompts for business that any team lead can copy, adapt, and use today. It is written for people who manage a team, own a workflow, or make decisions, not for developers or prompt engineers. Every prompt here is ready to run on ChatGPT, Claude, Copilot, or Gemini. Nothing requires a plugin, API access, or technical setup.
The pack contains 20 prompts across five categories: internal communications, project planning, hiring and HR, customer-facing writing, and reporting. Each prompt includes a plain-language label, a note on when to use it, and a clear pointer to the first variable you should change.
One thing this pack does not do well: it will not replace the judgment calls that require context only you have. If your situation is genuinely unusual, or the stakes are high enough that a wrong word costs a client relationship, treat any output as a first draft that needs a human pass, not a finished product.
For prompts organized by job function, see the full prompt library by role.
The prompts, by category
These 20 prompts are grouped into five categories: strategy, communications, operations, hiring, and finance. Each sits in its own copyable block with variables in brackets. For every prompt you get one line on when to reach for it and one line on what to change first.
Five prompts are flagged [NEEDS REAL OUTPUT]. Those sections are placeholders for captured model output; they will be replaced before publication.
If your team works primarily in marketing or sales, the AI Prompts for Marketers and AI Prompts for Sales Teams pages go deeper on those specific workflows. The prompts here are generalist by design.
One honest caveat before you start: these prompts work well for drafting and structuring, but they are not a substitute for human judgment on decisions with legal, financial, or personnel consequences. If a prompt output will be acted on directly rather than reviewed, slow down.
Strategy
Summarize a long document into a board-ready briefing
When to use it: Before an executive meeting when you need the key points from a lengthy report, contract, or research document. Change first: The audience line. "Board" implies brevity and financial framing; "department head" may need more operational detail.
VARIABLES: [DOCUMENT TEXT OR PASTE], [AUDIENCE], [DECISION TO BE MADE]
You are preparing a briefing for [AUDIENCE].
Below is a document they need to understand before making a decision about [DECISION TO BE MADE].
Summarize it in no more than 400 words using these sections:
- Situation (2-3 sentences)
- Key findings (bullet list, max 5 items)
- Recommended action (1-2 sentences)
- Open questions (bullet list, max 3 items)
Omit background that the audience already knows. Use plain language; avoid jargon.
Document:
[DOCUMENT TEXT OR PASTE][NEEDS REAL OUTPUT] What to notice: Whether the model correctly identifies the decision-relevant facts versus background noise.
Turn a messy goal into a 90-day plan
When to use it: At the start of a quarter when a stakeholder has given you a direction but not a plan. Change first: The constraints field. Budget, headcount, and dependencies change the output significantly.
VARIABLES: [GOAL], [TEAM SIZE], [CONSTRAINTS], [QUARTER START DATE]
I need a 90-day plan to achieve the following goal: [GOAL].
My team has [TEAM SIZE] people. Constraints include: [CONSTRAINTS].
The quarter starts on [QUARTER START DATE].
Structure the plan as three monthly phases. For each phase, list:
- The primary focus
- Two to four specific milestones with target completion dates
- One key risk and a mitigation step
Be specific. Do not include generic advice that applies to any goal.Write a SWOT analysis from a brief description
When to use it: Early in a planning cycle when you need a structured starting point for team discussion. Change first: The competitor context. A SWOT without named competitive pressures tends to produce generic output.
VARIABLES: [COMPANY OR PRODUCT DESCRIPTION], [TARGET MARKET], [MAIN COMPETITORS]
Write a SWOT analysis for the following:
Company or product: [COMPANY OR PRODUCT DESCRIPTION]
Target market: [TARGET MARKET]
Main competitors: [MAIN COMPETITORS]
Format as a 2x2 table with short bullet points in each cell. Limit each cell to four bullets. Prioritize the most consequential items, not an exhaustive list.Stress-test a business decision
When to use it: Before committing to a significant hire, contract, or strategic pivot. Change first: The decision description. Vague inputs produce vague challenges; be specific about what you're actually deciding.
VARIABLES: [DECISION], [SUPPORTING REASONING], [TIMEFRAME]
I am considering the following decision: [DECISION].
My current reasoning in favor: [SUPPORTING REASONING].
The timeframe for this decision: [TIMEFRAME].
Act as a skeptical senior advisor. Identify the three strongest arguments against this decision. For each, explain the specific risk and suggest one question I should be able to answer before proceeding.
Do not validate my reasoning. Challenge it.[NEEDS REAL OUTPUT] What to notice: Whether the model surfaces risks you hadn't named in the supporting reasoning, or simply restates your own concerns back at you.
Communications
Draft an internal announcement for a policy change
When to use it: When rolling out a new policy and you need a tone that is clear without being bureaucratic. Change first: The "what changes" field. Precision here prevents the model from softening or overgeneralizing the actual change.
VARIABLES: [POLICY NAME], [WHAT CHANGES], [EFFECTIVE DATE], [WHO IS AFFECTED], [WHERE TO ASK QUESTIONS]
Write an internal announcement for the following policy change.
Policy: [POLICY NAME]
What changes: [WHAT CHANGES]
Effective date: [EFFECTIVE DATE]
Affected employees: [WHO IS AFFECTED]
Contact for questions: [WHERE TO ASK QUESTIONS]
Tone: direct and plain. Do not use corporate filler phrases. Do not bury the key change. Lead with what is different, then explain why, then give the action step.
Length: under 200 words.Rewrite an email to reduce its length by half
When to use it: When a draft email is too long and you know it but can't see what to cut. Change first: The goal of the email. If the model doesn't know what the email needs to accomplish, it will cut the wrong things.
VARIABLES: [ORIGINAL EMAIL], [GOAL OF THE EMAIL]
Rewrite the following email so it is no more than half its current length.
The goal of the email is: [GOAL OF THE EMAIL].
Rules:
- Keep every piece of information the recipient needs to act
- Remove all pleasantries that do not affect the relationship
- Remove any background the recipient already knows
- End with a single, specific call to action
Original email:
[ORIGINAL EMAIL]Write a difficult message (declining, correcting, or ending something)
When to use it: When you need to say no, push back, or close something out without creating unnecessary friction. Change first: The relationship context. A message to a vendor is different from one to a colleague or a long-standing client.
VARIABLES: [SITUATION], [WHAT YOU NEED TO COMMUNICATE], [YOUR RELATIONSHIP TO THE RECIPIENT], [DESIRED OUTCOME]
Write a professional message for the following situation.
Situation: [SITUATION]
What I need to communicate: [WHAT YOU NEED TO COMMUNICATE]
My relationship to the recipient: [YOUR RELATIONSHIP TO THE RECIPIENT]
Desired outcome: [DESIRED OUTCOME]
Tone: direct but respectful. Do not over-apologize. Do not hedge to the point where the message becomes unclear. One paragraph maximum unless the situation requires more.[NEEDS REAL OUTPUT] What to notice: Whether the model balances clarity with tact, or defaults to excessive hedging that undermines the message.
Turn meeting notes into action items
When to use it: Immediately after a meeting to get a clean list of owners and deadlines from messy notes. Change first: The format. Some teams want a table; others want a short bulleted list they can paste into Slack.
VARIABLES: [RAW MEETING NOTES], [OUTPUT FORMAT: table or bullet list]
Below are raw notes from a meeting. Extract every action item.
For each action item, identify:
- What needs to be done
- Who is responsible (if named in the notes)
- The deadline (if mentioned; leave blank if not)
Output format: [OUTPUT FORMAT]
If the notes are ambiguous about ownership or timing, flag that item rather than guessing.
Notes:
[RAW MEETING NOTES]Operations
Write a standard operating procedure (SOP) from a verbal description
When to use it: When a process exists in someone's head and needs to be written down so others can follow it. Change first: The audience field. An SOP for a new hire reads differently from one for an experienced team member covering an edge case.
VARIABLES: [PROCESS NAME], [VERBAL DESCRIPTION OF STEPS], [AUDIENCE], [TOOLS USED]
Write a standard operating procedure for the following process.
Process name: [PROCESS NAME]
Description: [VERBAL DESCRIPTION OF STEPS]
Audience: [AUDIENCE]
Tools involved: [TOOLS USED]
Format:
1. Purpose (one sentence)
2. Scope (who this applies to)
3. Steps (numbered, each step as a single clear action)
4. Notes or exceptions (bullet list)
Use plain language. Assume the reader has not done this before.Identify the bottleneck in a described workflow
When to use it: When a process is slow or breaking down and you need a structured analysis before a team conversation. Change first: The symptoms field. The more specific the symptoms, the more targeted the analysis.
VARIABLES: [WORKFLOW DESCRIPTION], [SYMPTOMS OR COMPLAINTS], [TEAM SIZE AND ROLES INVOLVED]
Analyze the following workflow for bottlenecks.
Workflow: [WORKFLOW DESCRIPTION]
Symptoms observed: [SYMPTOMS OR COMPLAINTS]
Team involved: [TEAM SIZE AND ROLES INVOLVED]
Identify the most likely bottleneck. Explain your reasoning. Then suggest two specific changes, each with an implementation note and a potential downside.
Do not suggest adding more people as a solution unless you explain why that specifically addresses the bottleneck.[NEEDS REAL OUTPUT] What to notice: Whether the model correctly traces the bottleneck to a structural cause rather than defaulting to a generic suggestion like "improve communication."
Create a project status update from scattered notes
When to use it: When you need to send a stakeholder update and your notes are fragmented. Change first: The audience expectations. A status update for a sponsor who wants one number is different from one for an operational team.
VARIABLES: [RAW NOTES OR UPDATE FRAGMENTS], [PROJECT NAME], [AUDIENCE], [REPORTING PERIOD]
Write a project status update for [PROJECT NAME] covering [REPORTING PERIOD].
Audience: [AUDIENCE]
Raw notes: [RAW NOTES OR UPDATE FRAGMENTS]
Format:
- Overall status: Red / Amber / Green (choose based on the notes, explain the choice)
- Completed this period (bullet list)
- In progress (bullet list with % complete where known)
- Blockers (bullet list; flag if owner action is needed)
- Next steps (bullet list with dates)
Be specific. Do not pad with positive framing that contradicts the status rating.Write a vendor evaluation scorecard
When to use it: Before comparing proposals or shortlisting suppliers, to align the team on what actually matters. Change first: The criteria. Generic criteria produce generic scorecards; the value is in weighting factors that reflect your specific priorities.
VARIABLES: [PURCHASE CATEGORY], [MUST-HAVE REQUIREMENTS], [NICE-TO-HAVE REQUIREMENTS], [TEAM SIZE USING THIS], [BUDGET RANGE]
Create a vendor evaluation scorecard for purchasing [PURCHASE CATEGORY].
Must-have requirements: [MUST-HAVE REQUIREMENTS]
Nice-to-have requirements: [NICE-TO-HAVE REQUIREMENTS]
Team size: [TEAM SIZE USING THIS]
Budget range: [BUDGET RANGE]
Output a table with:
- Column 1: Criterion
- Column 2: Weight (% of total, all weights sum to 100)
- Column 3: Scoring guide (what a 1, 3, and 5 score looks like for this criterion)
- Column 4: Vendor A score (leave blank)
- Column 5: Vendor B score (leave blank)
Include both must-have and nice-to-have criteria, weighted accordingly.Hiring
Write a job description from a role brief
When to use it: When you know what you need from a hire but haven't written it up formally. Change first: The "what success looks like" field. This is what separates a useful job description from a generic one.
VARIABLES: [JOB TITLE], [TEAM AND REPORTING LINE], [CORE RESPONSIBILITIES], [WHAT SUCCESS LOOKS LIKE AT 6 MONTHS], [MUST-HAVE SKILLS], [NICE-TO-HAVE SKILLS], [LOCATION AND WORK ARRANGEMENT]
Write a job description for a [JOB TITLE] role.
Team and reporting line: [TEAM AND REPORTING LINE]
Core responsibilities: [CORE RESPONSIBILITIES]
Success at 6 months: [WHAT SUCCESS LOOKS LIKE AT 6 MONTHS]
Must-have skills: [MUST-HAVE SKILLS]
Nice-to-have skills: [NICE-TO-HAVE SKILLS]
Location and work arrangement: [LOCATION AND WORK ARRANGEMENT]
Format: brief intro paragraph, responsibilities as a bullet list, requirements as two separate bullet lists (must-have and nice-to-have). No salary range unless provided. Avoid superlatives like "rockstar" or "ninja."Write structured interview questions for a specific role
When to use it: Before an interview panel to ensure all interviewers are assessing the same competencies. Change first: The competencies. Each question should map to one competency; if the competencies aren't right, the questions won't be either.
VARIABLES: [JOB TITLE], [FOUR KEY COMPETENCIES], [SENIORITY LEVEL]
Write structured behavioral interview questions for a [SENIORITY LEVEL] [JOB TITLE].
The four competencies to assess: [FOUR KEY COMPETENCIES]
For each competency, write:
- One behavioral question (starting with "Tell me about a time...")
- One follow-up probe question
- Two to three indicators of a strong answer
Format as a table: Competency | Question | Follow-up | Strong answer indicators.Write an offer letter covering letter (not the legal document)
When to use it: When you want to send a warm, personal note alongside the formal offer to improve acceptance rate. Change first: The specific thing you are most excited about in this candidate. Generic enthusiasm is easily spotted.
VARIABLES: [CANDIDATE NAME], [ROLE], [WHAT STOOD OUT ABOUT THIS CANDIDATE], [ONE THING ABOUT THE TEAM OR MISSION RELEVANT TO THEM], [START DATE]
Write a brief, warm cover note to accompany a job offer to [CANDIDATE NAME] for the role of [ROLE].
What stood out about this candidate: [WHAT STOOD OUT ABOUT THIS CANDIDATE]
Relevant team or mission detail: [ONE THING ABOUT THE TEAM OR MISSION RELEVANT TO THEM]
Proposed start date: [START DATE]
Tone: genuine, not effusive. Two short paragraphs maximum. Do not use phrases like "we are thrilled" or "we are excited to offer." Say something specific about why this person, for this role.Finance and reporting
Turn raw financial data into a plain-English summary
When to use it: When you need to brief a non-finance audience on financial results without losing accuracy. Change first: The audience assumptions. What a department head needs to know differs from what a board observer needs to know.
VARIABLES: [FINANCIAL DATA OR TABLE], [REPORTING PERIOD], [AUDIENCE AND THEIR FINANCIAL LITERACY], [TWO OR THREE KEY QUESTIONS THEY WILL ASK]
Summarize the following financial data in plain English for [AUDIENCE AND THEIR FINANCIAL LITERACY].
Reporting period: [REPORTING PERIOD]
Key questions they will likely ask: [TWO OR THREE KEY QUESTIONS THEY WILL ASK]
Data:
[FINANCIAL DATA OR TABLE]
Write three sections:
1. What happened (two to three sentences, factual)
2. What drove it (the two or three most significant factors, each in one sentence)
3. What it means for next steps (one to two sentences, forward-looking)
Do not use accounting jargon without defining it. Do not editorialize beyond what the data supports.Write a budget justification for a new initiative
When to use it: When you need to request budget and want to pre-empt the most obvious objections. Change first: The objections field. Anticipating the right objections requires knowing your approver, not just the initiative.
VARIABLES: [INITIATIVE NAME], [REQUESTED BUDGET], [WHAT THE BUDGET COVERS], [EXPECTED RETURN OR BENEFIT], [TIMELINE
## Real example outputs
Five of the prompts above have been run against real models. The outputs are marked [NEEDS REAL OUTPUT] and will be replaced with captured responses before publication. Until then, here is what to watch for in each case.
| Prompt | Model tested | What a good output looks like | Common failure mode |
|---|---|---|---|
| Weekly status report | GPT-4o | Bullet summary under 200 words, no filler phrases | Adds a "challenges" section you did not ask for |
| Job description | Claude 3.5 Sonnet | Plain requirements list, no corporate buzzwords | Overloads "responsibilities" and buries the salary |
| Customer complaint reply | Copilot | Acknowledges the specific issue, proposes one fix | Generic apology with no resolution path |
| Competitor positioning | Gemini 1.5 Pro | Table with one honest weakness per competitor | Flatters your product, softens rival strengths |
| Meeting agenda | GPT-4o | Times assigned to each item, owner named | Lists topics without time blocks, useless for facilitation |
:::keypoint heading="Worth pausing on"
Outputs vary by model. A prompt that produces a clean status report in GPT-4o may return a padded essay in Gemini. Test on the model your team actually uses before sharing a prompt widely.
:::
## How to adapt these for your team
The prompts above work as written, but they work better once you've shaped them to your context. The fastest way to do that is to replace the generic variables with real defaults: your company name, your tone-of-voice guide, your standard deal stages, your preferred output format.
:::keypoint heading="Worth pausing on"
A prompt your team uses daily should have your context baked in, not typed fresh each time. One edited, saved version beats twenty slightly different improvised ones.
:::
Start with the two or three prompts your team reaches for most. Add a standing instruction at the top, something like "We write in plain, direct English for mid-market operations managers. Avoid jargon." That single line shifts every output without requiring further editing.
If your team spans roles, split the library by function rather than keeping one flat list. [Prompts organized by role](/prompts) make it easier for each person to find what's relevant without scrolling past categories that don't apply to them.
One honest caveat: prompts tuned tightly to one context can produce outputs that feel too narrow when the brief changes. Keep an unedited copy of the original so you can reset quickly.
## Save this folder to your workspace
These prompts work best when your team can find and reuse them without digging through chat history. A shared folder keeps the versions your team has actually tested, with the variables already filled in for your context.
:::cta heading="Want your team using the same prompts every time?" href="/product/shared-folders" button="Start free"
Convergence shared folders let you store, organize, and push prompts across ChatGPT, Claude, Copilot, and Gemini, so everyone starts from the same tested version.
:::
A few things worth doing before you save:
- Replace the generic [BRACKETS] with your actual defaults. If your company always targets mid-market SaaS buyers, write that in once.
- Add a note to each prompt naming the last person who improved it and what they changed. Version drift is how prompt libraries go stale.
- Archive prompts that are no longer used rather than deleting them. Context about what didn't work is genuinely useful.
For prompts organized by the role using them, the [full prompt library by role](/prompts) is a good starting point for building out other folders.
## Frequently asked questions
### Do these prompts work on Claude and Gemini, or only ChatGPT?
All the prompts on this page work across ChatGPT, Claude, Copilot, and Gemini without modification. The variables in brackets keep the structure consistent regardless of which model you use. If one model returns an output that feels too formal or too brief, adjust the tone instruction before swapping platforms.
### What if my team has no AI experience?
Start with the meeting summary and status update prompts. They have the shortest learning curve and produce useful output on the first try.
### Are these prompts safe to use with confidential business information?
Check your organization's data policy before pasting internal documents into any AI tool. Replacing real names and figures with placeholders is a reliable habit until your IT team confirms what is permitted.
### When should I not use a prompt template?
Skip templates when the task is genuinely one-off. A prompt built for repeatable work adds friction to something you will only do once.
### How do I share these prompts with my whole team?
[Convergence shared folders](/product/shared-folders) let you store, organize, and distribute prompts to everyone on your team from one place.