50 AI Prompt Examples You Can Copy, Paste, and Improve

A copy-paste-worthy AI prompt does three things a throwaway one never does: it assigns a clear role, it hands over the specific context the model can't guess, and it defines exactly what the output should look like. The difference between "write me a cover letter" and a prompt that produces something you'd actually send is rarely talent — it's structure. Below are 50 prompts you can paste into any chatbot today, grouped by job, each one written to be specific enough to use and clear enough to adapt.

Before you scroll: every prompt here has a bracketed placeholder like [paste your draft] or [your role]. Those brackets are the whole game. Fill them with real, concrete detail and the output sharpens immediately. Leave them vague and you get the generic mush that makes people think AI "doesn't really work." If you want the deeper logic behind why these are shaped the way they are, how to prompt AI correctly and the RCTCO prompt structure cover it. For now, let's get tactical.

What makes a prompt worth copying

Three traits separate a prompt you can lift and reuse from one that wastes a turn.

It removes ambiguity. A weak prompt leaves the model guessing your audience, format, length, and standard of quality. A strong one states them. The model isn't being lazy when it gives you a bland answer — it's hedging across every interpretation of a vague request.

Weak: Help me write a LinkedIn post about our launch.

Strong: Write a 120-word LinkedIn post announcing the launch of our budgeting app for freelancers. Audience: solo designers and writers. Tone: confident, not hypey. Open with a pain point, end with one question to drive comments. No hashtags.

It encodes the context the model can't infer. Your product name, your audience, the constraint that matters (word count, reading level, deadline, brand voice). The model knows everything about the world in general and nothing about your specific situation — that's the gap a good prompt closes.

It defines the output shape. "A table with three columns," "five bullet points," "a 200-word summary at a 9th-grade reading level." When you don't specify the container, you get whatever the model defaults to, which is usually a wall of prose. Keep these three traits in mind as you read — every prompt below is built on them, and that's what makes them adaptable rather than disposable.

Writing and editing prompts

Rewrite the following text to be 30% shorter without losing any key point. Keep the tone professional but warm, and use plain language a non-expert could follow: [paste your draft]

You are a line editor. Edit the paragraph below for clarity, rhythm, and concision. Show the edited version first, then a short bulleted list of what you changed and why: [paste paragraph]

Generate 10 headline options for an article titled "[working title]." Make 3 curiosity-driven, 3 benefit-driven, 3 how-to style, and 1 contrarian. Keep each under 65 characters.

Rewrite this email to sound more confident and less apologetic. Remove hedging words like "just," "I think," and "sorry to bother." Keep it under 90 words: [paste email]

Take the bullet points below and turn them into a smooth, flowing 3-paragraph narrative for a company blog. Keep every fact, add no new claims: [paste bullets]

Act as a skeptical reader. Read my argument below and list the 5 weakest points or unsupported claims, then suggest how I'd strengthen each one: [paste argument]

Continue this story in the same voice and tense for about 200 words, ending on a moment of tension: [paste opening]

Why these work: each names a target (shorter by 30%, under 65 characters, 200 words) and a transformation. You're not asking the model to "improve" something — a word it can't measure — you're asking for a specific, checkable change.

Work and email prompts

Draft a reply to the email below. I want to decline the request politely but keep the relationship warm, and propose an alternative time in two weeks. Keep it under 100 words: [paste email]

Summarize the meeting notes below into: (1) decisions made, (2) action items with owners, (3) open questions. Use bullet points and bold each owner's name: [paste notes]

Turn this rough brain-dump into a clear, professional Slack message to my team. Lead with the ask, then context, then deadline: [paste brain-dump]

Write three subject lines and a 75-word follow-up email to a prospect who went quiet after a demo. Be friendly, give them an easy out, and include one new reason to re-engage.

I need to give difficult feedback to a teammate who keeps missing deadlines. Draft what I could say in a 1:1 — direct but supportive, specific about impact, and ending with a question that invites their side.

Convert the project update below into a 5-line status report for executives. Lead with the headline (on track / at risk / blocked), then the why, then what I need from them: [paste update]

Write an out-of-office message for [dates] that's professional, sets expectations on response time, and names [colleague] as the contact for urgent issues.

Why these work: they specify the recipient and the relationship — exec, prospect, teammate — so the model can calibrate tone and length. Audience is the single most underused variable in workplace prompts.

Coding prompts

You are a senior [language] engineer. Review the function below for bugs, edge cases, and readability. List issues by severity, then show a corrected version: [paste code]

Explain what this code does line by line, as if to a junior developer. Then note one thing that could break it in production: [paste code]

Write a [language] function that [describe the task]. Include input validation, handle the empty and null cases, and add a short docstring with one usage example.

Convert this code from [language A] to [language B]. Keep the logic identical, use idiomatic patterns for the target language, and flag anything that doesn't translate cleanly: [paste code]

I'm getting this error: [paste error and stack trace]. Here's the relevant code: [paste code]. Walk me through the most likely cause, then the fix, then how to prevent it next time.

Write a set of unit tests for the function below. Cover the happy path, boundary values, and at least two failure cases. Use [test framework]: [paste function]

Refactor this code to be more readable without changing its behavior. Explain each change in one line so I can learn the reasoning: [paste code]

Why these work: they ask the model to reason before it writes — review, then correct; diagnose, then fix. For anything technical, asking for the reasoning step first dramatically cuts silent mistakes, which is the core idea behind chain-of-thought prompting.

Research and analysis prompts

Summarize the article below in 150 words, then list the 3 claims that would most need fact-checking before I cite them: [paste text]

Compare [option A] and [option B] across cost, speed, ease of use, and long-term flexibility. Present it as a table, then give a one-paragraph recommendation for someone who values [priority].

I'm researching [topic]. Give me the 5 questions a true expert would ask that a beginner wouldn't think of, and briefly explain why each matters.

Read the data below and tell me the 3 most interesting patterns or anomalies. For each, state your confidence and what additional data would confirm it: [paste data]

Act as a devil's advocate against this plan. Give me the strongest case for why it might fail, then rate each risk as low, medium, or high likelihood: [paste plan]

Break down the concept of [topic] into a one-paragraph executive summary, then a deeper section for someone who needs to actually implement it.

Extract every named entity (people, companies, dates, dollar amounts) from the text below and organize them into a table by category: [paste text]

Why these work: they force the model to qualify itself — confidence levels, what would confirm a finding, what needs fact-checking. That single move turns confident-sounding output into something you can actually trust, because it surfaces uncertainty instead of hiding it.

Learning and study prompts

Explain [concept] to me at three levels: like I'm 10, like I'm a high schooler, and like I'm a graduate student. Use a different analogy for each.

I want to learn [skill] in 4 weeks, studying 5 hours a week. Build me a week-by-week plan with specific milestones and one project to prove I've learned it.

Quiz me on [topic]. Ask one question at a time, wait for my answer, tell me if I'm right, and adjust the difficulty based on how I do. Start now.

I just read this passage but I'm not sure I understood it. Ask me 3 questions to test my comprehension, then fill in whatever I get wrong: [paste passage]

Turn the notes below into a set of 10 flashcards in question-and-answer format, ordered from foundational to advanced: [paste notes]

Teach me [topic] using the Feynman technique: explain it simply, then point out exactly where my understanding would likely have gaps.

I keep confusing [concept A] and [concept B]. Give me a memorable way to tell them apart and one example where the distinction actually matters.

Why these work: the best learning prompts make the model interactive and adaptive — quiz me, wait, adjust — instead of dumping information. They also anchor to your level and timeframe, so the answer fits you rather than an average learner. Students will find more of these in our ChatGPT prompts for students guide.

Personal and life admin prompts

Help me plan a 3-day trip to [city] for someone who likes [interests] and wants to avoid tourist traps. Give a day-by-day itinerary with one backup option per day in case of rain.

I have these ingredients: [list]. Suggest 3 dinners I could make tonight, ranked by how little extra I'd need to buy. Include rough cook times.

Draft a polite but firm message to my landlord about [issue]. State the problem, reference our lease if relevant, and request a specific resolution by a specific date.

Help me write a 6-line toast for [occasion] for [person]. Warm, a little funny, one specific memory — not generic.

I'm overwhelmed by my to-do list below. Help me sort it into "do today," "schedule," "delegate," and "drop," and tell me which single item to start with: [paste list]

Create a simple weekly meal plan for [number] people that's [dietary constraint], stays under [budget], and reuses ingredients across meals to cut waste.

Why these work: they bring real constraints — budget, dietary needs, weather backups, a specific person — into the request. Personal prompts fail when they're generic; they shine the moment you hand the model the messy specifics of your actual life.

Business and marketing prompts

Write 5 ad headlines and 2 descriptions for [product] targeting [audience]. Each headline under 30 characters, each description under 90. Lead with the benefit, not the feature.

I'm launching [product]. Draft a 5-email welcome sequence: email 1 welcome, 2 the core benefit, 3 a use case, 4 social proof, 5 a soft offer. One paragraph each, with subject lines.

Take the product feature below and rewrite it as a customer-facing benefit. Then write the one-sentence version I'd put on a landing page: [paste feature]

Generate 15 content ideas for [brand] aimed at [audience], split across educational, behind-the-scenes, and conversion-focused. Format as a table with the angle and the hook for each.

Write a 280-character post announcing [news] in our brand voice: [describe voice]. Make it specific, skippable-resistant, and end with a clear reason to click.

Draft a customer-facing apology for [issue]. Take responsibility without over-promising, explain what we're doing about it, and offer a concrete next step. Keep it under 120 words.

Act as a positioning strategist. Given my product description below, give me 3 distinct positioning angles, each with the audience it targets and the message that lands: [paste description]

Why these work: marketing prompts live or die on audience and constraint. Every one above names who it's for and caps the format, so you get usable copy instead of a vague essay about marketing. For a deeper set, see our ChatGPT prompts for marketing collection.

How to adapt any of these

These 50 are starting points, not finished tools. The fastest way to make one yours is to layer in the three traits from the top: a sharper role, your real context, and a tighter output spec.

Take any prompt and run it through three quick passes. First, swap the role to match your standard — "a senior copywriter," "a skeptical CFO," "a patient tutor" — because the role sets the depth and vocabulary before the task even starts. If that idea is new to you, role prompting explains why it moves the needle so much.

Second, load the real context. Replace every bracket with specifics, and add the one detail you keep in your head but didn't write down — the audience's objection, the brand quirk, the constraint your boss cares about. That's almost always the missing ingredient between a generic answer and a great one.

Third, tighten the output. Add a length, a format, a reading level, or an example of what "good" looks like. Showing the model one or two examples of the style you want — few-shot prompting — is the single biggest upgrade you can make to any prompt here.

If you'd rather build prompts visually instead of editing text in your head, Studio lets you assemble them from blocks — Goal, Role, Context, Instruction — and version them as you improve. And if you just want a great answer fast, paste your rough request into the free brain at the top of the site: type something messy, and it returns an expert-grade version you can use or learn from. No setup, no key.

Frequently asked questions

What is a good example of an AI prompt?

A good prompt names a role, supplies specific context, and defines the output. For example: "You are a senior editor. Rewrite the paragraph below to be 30% shorter, keep the tone warm, and list what you changed: [paste text]." It's checkable, specific, and leaves the model nothing important to guess. Any of the 50 above follow that pattern.

How do I write my own prompt instead of copying one?

Start from the structure, not a blank page. State who the AI should be, give it the context it can't know, describe the task, and specify the format and length you want. The RCTCO structure and our AI prompt template walk through this fill-in-the-blanks approach step by step.

Why do my prompts give worse results than these examples?

Usually it's missing specificity — no audience, no length, no format, no context the model could anchor to. Vague prompts force the model to average across every interpretation, which produces bland output. Add constraints and real detail and the quality jumps immediately. We unpack the common failure modes in why most AI results are mediocre.

Do these prompts work in ChatGPT, Claude, and Gemini?

Yes. The principles — role, context, task, format — work across every major model. You may tweak phrasing slightly for each, but a well-structured prompt is portable. If you're curious how the big models differ in how they respond, Claude vs ChatGPT prompts breaks it down.

Copy the ones you need, fill in the brackets, and run them. When you want to go further, browse the Library — 2,750+ expert prompts you can copy or open straight into Studio — and use the free brain to turn any half-formed idea into a prompt worth keeping.