A Working Prompt Playbook
ChatGPT Prompts for Business: The 12 I Hand Every Client
The most useful ChatGPT prompts for business are short briefs, not magic phrases: each one gives the tool your context, a role to work from, your source material, and a format for the answer. Below are the twelve prompts I hand coaching clients, organized by the job you need done.
1-on-1 with Pete Enestrom, on your actual work
Written by Pete Enestrom
Yale & Columbia, ex-Microsoft & Intel — 1-on-1 AI coaching for executives
Your coach’s background
This is not another list of two hundred prompts you will never open again. It is the opposite: the twelve ChatGPT prompts for business that I actually hand to the founders and executives I coach, organized by the job you need done—drafting, analysis, meetings, and decisions. Each one is copy-paste ready, with [brackets] where your own material goes.
Most AI prompts for business fail for a predictable reason: they are all task and no context. Ask ChatGPT to “write a follow-up email” and you will get a grammatically perfect email that sounds like nobody at your company, because ChatGPT knows nothing about your company. Ask with the situation, the relationship, and the outcome you want pasted in, and the same tool produces something you would actually send. Clients ask me for the best AI prompts for business as though a secret list exists somewhere. No list does. What exists is a pattern—context, role, format—and the twelve prompts below demonstrate that pattern on real executive work. Learn the pattern and you will never need another list.
Copy any prompt verbatim, replace the [brackets] with your material, and adjust the last line to taste. Then read the note under each card—the why-it-works teaching is what lets you write the thirteenth prompt yourself.
What separates a prompt that works from one that flops?
Every prompt that earns its keep has three ingredients, and you can check any prompt—yours or one from this page—against them. Context: the situation, the relationship, and the source material, pasted in, because ChatGPT cannot read your desk and the [brackets] are where your desk goes. Role: one line about the seat ChatGPT should answer from—“I am the tenant in this deal” or “do not validate me”—because an assistant with no assigned seat gives advice with no point of view. Format: what the answer should look like, in numbers—three sentences, a table, under 150 words, no corporate phrases. A one-line prompt supplies none of the three, so ChatGPT guesses at all of them, and the guess is always the generic middle.
- Context: who you are, the situation, and the actual material, pasted into the brackets
- Role: the seat ChatGPT should answer from—your side of the deal, a skeptic, the person across the table
- Format: the shape of a good answer—length, structure, tone, and the phrases it may not use
Drafting: emails, announcements, and your own voice
Writing jobs are where these tools hand back the most hours. Replace the brackets with your real situation—the relationship and the history are what make the drafts sound like your desk instead of a template.
Two weeks ago I sent a proposal to [type of client] for [the work] and haven't heard back. Draft a follow-up email: one sentence acknowledging they're busy, one sentence restating the outcome they said they wanted, and one short question they can answer with a single word. Under 120 words. Plain and warm—no “just checking in.”
Why it works: the format does the heavy lifting—three sentences, each with a job. The banned phrase stops ChatGPT from writing the follow-up everyone ignores.
I need to announce [the change: a price increase / a policy change / a reorganization] to [employees or customers]. Context: [two or three sentences on why this is happening]. Draft the announcement: state the change in the first sentence, give the honest reason in two, acknowledge what it costs the reader, and say exactly what happens next and when. About 200 words. Steady and human, not legal.
Why it works: context is the ingredient here. Without “why this is happening,” ChatGPT writes a notice; with it, ChatGPT writes an explanation a person would sign.
Here are three emails I have actually sent: [paste them]. Study the style—sentence length, tone, the phrases I reach for. Then draft [the email you need] about [the situation in two lines] in that style. Afterward, flag any sentence that sounds like generic AI writing and offer a plainer version.
Why it works: you supplied source material to imitate. The self-check instruction—“flag anything generic”—gives ChatGPT a quality bar to grade its own draft against.
Analysis: numbers, contracts, and reports
The ChatGPT business prompts for analysis share one rule: paste the actual document—the numbers, the contract, the quotes, the report. An analysis prompt with no source material attached is a horoscope. These four assume you will paste the real thing.
Here is our [P&L / sales export] for [this period]: [paste]. I am the [owner] of [one line about the business]. Identify the three numbers that moved most versus [the prior period], the most likely business explanation for each, and what you would examine next. If the data does not support a conclusion, say so instead of guessing.
Why it works: the role line—“I am the owner of a company that…”—aims the analysis at your decisions, and permission to say “I don't know” is what makes the answer trustworthy.
Below is [a lease renewal / a vendor agreement]: [paste]. I am the [tenant / customer] in this deal. List the clauses that most affect me, anything unusual compared with standard terms, and five questions to bring to my attorney before I sign. Plain English—no legalese back at me.
Why it works: the role (“I am the tenant”) tells ChatGPT whose side to read from. The output organizes your thinking so the paid hour with your attorney goes further—it does not replace the attorney.
I am choosing between [vendor A] and [vendor B] for [the work]. Here is what each quoted: [paste both proposals]. Build a comparison table covering total cost, scope differences, what each excludes, and risk for a company of [your size]. Then list the three questions whose answers would most change the decision.
Why it works: you named the criteria instead of accepting ChatGPT's, and the table format forces side-by-side honesty where prose would blur the differences.
Here is [an industry report or article]: [paste]. I run [one line about your company]. Tell me what in this report should change what I do in the next two quarters, and what is noise for a company my size. Be specific about my situation rather than the industry in general.
Why it works: “for a company my size” is the context line that turns generic industry commentary into a decision aid. Without it, you get the same summary everyone else got.
Meetings: agendas, pushback, and interviews
Meetings reward preparation and punish winging it. These three AI business prompts cover the meeting lifecycle: the agenda before, the rehearsal for the hard one, and the interview you are running.
I lead a [weekly leadership meeting] with [number] people. This week we must cover [the topics]. Draft a [45]-minute agenda that puts decisions before updates, gives every item an owner and a timebox, flags anything that could have been an email, and reserves the last five minutes for open issues.
Why it works: “decisions before updates” is a management philosophy delivered as a format instruction. The prompt encodes how you run meetings, not just what this one covers.
In tomorrow's meeting, [who] will push for [the plan or demand]. I am skeptical because [your reasons in two lines]. Play the strongest, most reasonable version of their side and argue it well. Then give me the three weakest points in that argument and one question that exposes each without putting anyone on the defensive.
Why it works: the role instruction—“play their side”—is doing everything. Arguing the other case first is rehearsal no busy colleague has time to give you.
I am interviewing a [role] candidate tomorrow. The job in one line: [what this person will own]. The resume: [paste]. Give me five questions that probe the gaps between the resume and the job, two follow-ups for the question where I am most likely to accept a vague answer, and one red flag to watch for.
Why it works: all three ingredients appear—context (the job), source material (the resume), and format (questions, follow-ups, red flag). Steal the follow-up instruction for any interview.
Decisions: second opinions and pre-mortems
The highest-value prompts do not produce documents; they produce better thinking. These two are the ones clients keep longest, because a second opinion that argues back is the thing a busy owner cannot buy easily.
Here is a decision I am about to make: [the decision, the options, which way I am leaning, and why]. Do not validate me. Make the strongest case that I am wrong, list what I would have to believe for my choice to be right, and name the one thing you would verify before committing.
Why it works: left alone, these tools agree with you. “Do not validate me” reassigns the role from cheerleader to sparring partner—and the sparring partner is the useful one.
We are about to [launch the product / sign the lease / make the hire]: [two or three sentences of detail]. Imagine it is six months from now and it failed. Write the failure story: the three most likely causes in order, the early warning sign we ignored for each, and what we could change this week to prevent it.
Why it works: the imagined future is a role-play that unlocks candor—people and models both spot risks more honestly when failure is the premise rather than the fear.
How do you turn a borrowed prompt into your own?
The twelve cards above are starting points. Four moves turn any of them—or any prompt you find anywhere—into something that fits your business.
Fill every bracket with real material
Paste the actual email, the real numbers, the genuine proposal. A prompt is twenty percent of the output quality; the material you attach is the other eighty. If one bracket feels like effort, that bracket is the one that matters.
Add one line of company context
Who you are, how big you are, and what a good outcome looks like, in a single sentence. “I run a forty-person commercial HVAC firm” changes every answer that follows it.
Pin down the format before you send
Length, structure, tone, and banned phrases—tell ChatGPT what “done” looks like. “Under 150 words, three bullet points, no exclamation marks” is the difference between editing a draft and rewriting one.
Save the keepers and reuse them
When a prompt produces something you would sign, save the exact wording in a note and run it again next week. A dozen prompts reused monthly beat two hundred prompts bookmarked once. That note is how this list becomes your playbook.
Why does the same prompt work for one owner and flop for another?
I watch this in coaching sessions constantly. Two owners run the identical prompt; one gets a draft worth sending, the other gets mush. The difference is never intelligence and rarely the tool—it is five habits, all fixable in a week.
- Vague asks. “Help me grow my business” has no context, no role, and no format, so the answer has no use.
- No source material. Asking ChatGPT to summarize a document you did not paste invites invention.
- Skipping the role line. Whose side ChatGPT is on is what gives the advice a point of view.
- Accepting the first draft. The reply—“shorter, plainer, and the second point is wrong”—is where the quality lives.
- Pasting sensitive material without a policy. Strip names and figures until your company has written rules for what may be pasted.
What is the difference between a prompt list and prompt fluency?
A list gives you a draft today; fluency gives you every draft after that. The executives I coach count as fluent when they stop collecting prompts and start composing them—opening a blank chat, stating the context, assigning the role, specifying the format, and iterating twice before judging the output. At that point the tool stops being a novelty and starts being Tuesday.
That composing skill is what I teach in 1-on-1 coaching sessions over video, with founders and executives across the US: we open your real workload and build your prompt playbook on your documents, your voice, and your risk tolerance. This page is the starting twelve; coaching is how the thirteenth prompt becomes yours.
1-on-1 coaching
Want a teacher, not another tutorial?
Guides get you started. A 1-on-1 session gets you fluent—built on your actual work, at your pace.
Your coach
Coached by Pete Enestrom
Yale and Columbia grad, former Microsoft and Intel, and a venture-backed exited founder. Pete has spent the last four and a half years going deep on every major AI tool, and he teaches the way operators learn: on your real work, at your pace, with nothing assumed.
Common questions
Straight answers, the way I'd give them across a table.
Keep reading
Resource
AI Prompt Library: What It Is and Why Every Executive Should Keep One
An AI prompt library is your personal collection of tested prompts, organized by the work you repeat. Why every executive should keep one, plus a starter set.
ReadGuide
AI Prompt Examples: How to Write Prompts That Actually Work
How to write AI prompts that work on real executive tasks: the four parts of a good prompt, before-and-after rewrites, and copy-paste examples you can use today.
ReadTraining
AI Fluency: What It Is and How Executives Get There
AI fluency defined in plain English: what fluent AI use looks like on an executive's real work, why certificates don't create fluency, and how leaders build it.
ReadGuide
ChatGPT for Work: A Practical Guide for Non-Technical Executives
How non-technical executives use ChatGPT at work: drafting, summarizing, brainstorming, notes into action lists, and meeting prep—with six copy-paste prompts.
ReadBeyond coaching
When the workflow needs to be encoded, not just learned — that's what our team at Zaigo builds.