AI Prompts

Welcome to your shortcut to better AI output.
A good prompt is the difference between a generic, forgettable answer and one that actually saves you time. This page is a working library of prompts for ChatGPT, Claude, Gemini, and image generators like Midjourney and DALL-E — organized by what you’re actually trying to do, not by which tool you happen to be using.

Every prompt here is written to be copied, pasted, and adapted. Fill in the bracketed placeholders with your own details and it works in any of the major AI chat tools.

Hit the 📋 Copy Prompt button on any card to copy it straight to your clipboard — no manual text-selecting needed. And before you dive in, the section below explains the one habit that separates these prompts from a typical “50 ChatGPT prompts” list.

The Framework Every Prompt Below Uses

Most prompt lists are one long run-on sentence, and that’s exactly why they produce generic, forgettable output. Every prompt in this library is instead built from four parts that professional prompt engineers actually use:

  • Role — who the AI should act as. This sets its judgment and vocabulary.
  • Task — the specific thing to produce, with your details filled in.
  • Format — exactly what the output should look like (a table, a word count, a specific structure), instead of leaving that to chance.
  • Constraints — the guardrails that stop the model from inventing facts, padding with fluff, or guessing at something it should ask about instead.

That last part is the one most people skip, and it’s the highest-leverage one. Several prompts below explicitly tell the AI to ask you a clarifying question instead of assuming — because a model that pauses to ask gives a more accurate result than one that guesses confidently and sounds right anyway. If you take one habit from this page and apply it to prompts you write yourself, make it that one.

Browse Prompts by Category

49 prompts across 8 categories — each with a short note on what it’s for and, where relevant, which AI tool tends to handle it best.

Writing & Content Prompts (7)

Prompts for drafting, rewriting, and polishing text — works well with ChatGPT, Claude, or Gemini interchangeably.

1. Blog Post Outline From a Topic

Turns a rough topic into a structured outline before you write a word.

Role: You are a senior content strategist who has outlined hundreds of high-performing articles.
Task: Build a complete outline for a blog post on [TOPIC], written for [AUDIENCE] in a [TONE] tone.
Format: A working title, a one-sentence hook for the intro, 5-7 H2 sections (specific to this topic, not generic labels like ‘Introduction’), each with 2-3 supporting bullet points, and one closing takeaway.
Constraints: If [TOPIC] is too broad to outline meaningfully, ask me one clarifying question before writing the outline.

2. Rewrite in a Different Tone

Same message, different voice — useful for adapting one piece of writing across audiences.

Role: You are a skilled editor who adapts writing for different audiences without losing meaning.
Task: Rewrite the text below in a [TONE — e.g. professional / casual / persuasive / friendly] tone.
Format: Roughly the same length as the original. Return only the rewritten text.
Constraints: Keep every fact, number, and key point unchanged. If a phrase could shift meaning depending on tone, keep it literal rather than guessing which way I meant it.

Text: [PASTE TEXT]

3. Punchier Headline Generator

Generates multiple headline options instead of settling for the first draft.

Role: You are a headline editor at a publication that A/B tests every headline before publishing.
Task: Generate 10 headline options for an article about [TOPIC].
Format: A numbered list mixing formats — at least one number-based headline, one question, one how-to, one curiosity-gap headline, and one benefit-driven headline. Each under 70 characters.
Constraints: No clickbait the article can’t actually deliver on. After the list, mark your single strongest pick and say why in one sentence.

4. Cold Outreach Email

A first-draft outreach email that avoids sounding like generic spam.

Role: You are someone who writes outreach emails people actually reply to, not delete.
Task: Write a short cold outreach email to [PERSON/ROLE] at [COMPANY]. The goal is [GOAL — e.g. a guest post, a partnership, a demo call].
Format: Under 120 words. Reference one specific, real detail about their work so it doesn’t read as a mass email. End with a low-friction call to action (not ‘let’s hop on a call’).
Constraints: No hype language (‘game-changing’, ‘excited to connect’). If I haven’t given you a real specific detail about them, ask me for one instead of inventing something that sounds plausible.

5. Social Media Caption Pack

One post, five caption options across different platforms and tones.

Role: You are a social media manager who writes differently for every platform instead of copy-pasting one caption everywhere.
Task: I’m posting about: [TOPIC/CONTENT]. Write 5 caption options.
Format: One for LinkedIn (professional, insight-led), one for Instagram (casual, short), one for X/Twitter (under 280 characters, punchy), one built around a question to drive comments, and one story-style caption. Add 3-5 relevant hashtags to each.
Constraints: Match the actual norms of each platform — don’t just reuse the same sentence with different hashtags stapled on.

6. Turn Rough Notes Into Polished Text

Cleans up messy bullet-point notes into a coherent paragraph or email.

Role: You are an editor who cleans up messy notes without changing what was actually said.
Task: Turn these rough notes into a clear, well-organized [paragraph / email / summary].
Format: Fix grammar and remove repetition. Match the length to what the notes actually contain — don’t pad it out.
Constraints: Don’t add information, examples, or conclusions that aren’t implied by the notes. If something is genuinely unclear, mark it as [CLARIFY: …] instead of guessing.

Notes: [PASTE NOTES]

7. Explain It Like I’m New to This (ELI5)

Simplifies dense or technical writing for a non-expert reader.

Role: You are a teacher who’s good at explaining hard things to total beginners.
Task: Explain the following in plain, simple language, using one short real-world analogy.
Format: Under 150 words. Define any unavoidable technical term in one clause right after using it.
Constraints: If the topic has an important nuance that a true ELI5 would hide, mention it briefly in one closing sentence rather than oversimplifying to the point of being wrong.

Topic/text: [PASTE TOPIC OR TEXT]

Business & Marketing Prompts (6)

Prompts for strategy documents, product copy, and marketing drafts that need a real structure, not just filler text.

1. SWOT Analysis Builder

Structures a strengths/weaknesses/opportunities/threats breakdown for a business, product, or idea.

Role: You are a business analyst who has run SWOT sessions for real companies, not a template generator.
Task: Do a SWOT analysis for [BUSINESS/PRODUCT/IDEA], based on: [CONTEXT/DETAILS].
Format: 3-4 specific points per category (Strengths, Weaknesses, Opportunities, Threats) — no vague statements like ‘strong brand’ without saying why it’s strong.
Constraints: Think through the four categories first, then write the final list. End with one clear strategic recommendation and the single biggest risk to that recommendation.

2. Product Description Writer

Turns a feature list into a description that sells the benefit, not just the spec.

Role: You are a copywriter who sells the outcome, not the spec sheet.
Task: Write a product description for [PRODUCT], aimed at [TARGET CUSTOMER], in a [TONE] tone.
Format: A 1-sentence hook, 3-4 bullet-point benefits (translate each feature into what it means for the customer), and a closing line with a call to action. Under 150 words total.
Constraints: Don’t invent a feature, spec, or claim I haven’t given you. If a benefit isn’t obvious from the info provided, ask rather than fabricate one.

3. Customer Persona Builder

Creates a usable customer persona from whatever you already know about your audience.

Role: You are a market researcher building a persona a marketing team can actually use, not a generic stock character.
Task: Based on this information about my customers: [PASTE DATA/DESCRIPTION], build one detailed customer persona for [PRODUCT/SERVICE CATEGORY].
Format: Name, age range, job/role, main goal, biggest frustration related to this category, where they spend time online, and one realistic objection they’d raise before buying.
Constraints: Base every detail on the data given — where you have to infer something, mark it as an assumption rather than stating it as fact.

4. Elevator Pitch Refiner

Tightens a rambling pitch into something that fits in 30 seconds.

Role: You are a pitch coach who cuts pitches down to what actually matters.
Task: Refine the raw pitch below into a 30-second elevator pitch (75-90 words).
Format: Problem, solution, who it’s for, and the one thing that makes it different — in that order, no jargon or buzzwords.
Constraints: If the raw pitch doesn’t actually state what makes it different, flag that gap instead of inventing a differentiator.

Raw pitch/notes: [PASTE]

5. Competitor Comparison Table

Structures a side-by-side comparison instead of a wall of comparison text.

Role: You are a product marketer who writes comparison content people trust because it isn’t one-sided.
Task: Create a comparison table for [YOUR PRODUCT/BRAND] vs [COMPETITOR 1] vs [COMPETITOR 2].
Format: A table with rows for price, core feature set, target user, and one clear weakness per product.
Constraints: Stay factual — explicitly name the one thing a competitor genuinely does better than you. A comparison with zero honest weaknesses reads as an ad, not a comparison.

Best for: See how AIInsider structures real comparisons in ChatGPT vs Claude vs Gemini vs Perplexity.

6. Ad Copy A/B Variations

Generates multiple ad angles to test instead of running with just one.

Role: You are a performance marketer who tests angles instead of guessing which one will work.
Task: Write 5 ad copy variations for [PRODUCT/OFFER] targeting [AUDIENCE] on [PLATFORM — e.g. Google, Meta, LinkedIn].
Format: One headline + 1-2 sentence body per variation, within that platform’s character limits. Vary the angle: benefit-led, pain-point-led, social-proof-led, urgency-led, question-led.
Constraints: Don’t reuse the same headline structure across variations — if two end up sounding the same, rewrite one.

Coding & Developer Prompts (6)

Prompts for everyday development work — explaining, debugging, testing, and cleaning up code.

1. Explain This Code

Get a plain-language walkthrough of unfamiliar code before you touch it.

Role: You are a senior engineer doing a code walkthrough for a teammate who didn’t write this code.
Task: Explain what the code below does, step by step, in plain language.
Format: Walk through it in the order it executes. Call out any part that looks fragile, non-obvious, or likely to cause a bug, and explain why.
Constraints: If a variable or function name makes its purpose genuinely unclear, say so rather than guessing its intent.

“`
[PASTE CODE]
“`

Best for: For which AI is actually better at coding tasks, see Claude vs ChatGPT: Why I Switched.

2. Debug This Error

Structured debugging instead of pasting an error and hoping for a lucky guess.

Role: You are a debugging partner who reasons through root causes instead of pattern-matching to the first guess.
Task: I’m getting this error: [PASTE ERROR MESSAGE]. Here’s the relevant code:
“`
[PASTE CODE]
“`
Format: First, reason through the most likely root cause in 2-3 sentences. Then give the corrected code with a one-line comment on what changed and why.
Constraints: If the error could plausibly have more than one root cause, name the second-most-likely one too rather than presenting only one guess as certain.

3. Write Unit Tests

Generates a starting test suite for a function instead of writing boilerplate by hand.

Role: You are a QA-minded engineer who writes tests that actually catch bugs, not tests that just pass.
Task: Write unit tests for the function below using [TESTING FRAMEWORK — e.g. pytest, Jest].
Format: Cover the normal case, at least two edge cases, and one case that should raise an error or return null/undefined. Name each test so its purpose is obvious from the name alone.
Constraints: If the function’s expected behavior on an edge case is genuinely ambiguous from the code alone, ask before assuming.

“`
[PASTE FUNCTION]
“`

4. Refactor for Readability

Cleans up working-but-messy code without changing what it does.

Role: You are a code reviewer focused on maintainability, not cleverness.
Task: Refactor the code below for readability without changing its behavior.
Format: Improve variable names, reduce nesting where reasonable, add comments only where the logic genuinely isn’t self-explanatory. After the code, list what you changed and why, as bullet points.
Constraints: Do not change any actual logic or output — if a change would alter behavior, flag it separately instead of silently including it.

“`
[PASTE CODE]
“`

5. Convert Code Between Languages

A working first-pass translation between languages/frameworks that still needs a human review.

Role: You are a polyglot engineer who translates idiomatically, not line-by-line.
Task: Convert this [SOURCE LANGUAGE] code to [TARGET LANGUAGE], keeping the logic identical.
Format: Use idiomatic patterns for the target language. After the code, list any part where a direct equivalent doesn’t exist and explain the workaround.
Constraints: If a library or built-in used in the source has no direct match in the target language, say so explicitly rather than silently picking an approximate substitute.

“`
[PASTE CODE]
“`

6. Generate a Regex Pattern

Turns a plain-English matching rule into a tested regex with an explanation.

Role: You are an engineer who tests a regex before shipping it, not one who guesses and hopes.
Task: Write a regular expression that matches: [DESCRIBE WHAT IT SHOULD MATCH, e.g. ‘Indian mobile numbers with optional +91 prefix’].
Format: The pattern, a one-line explanation of each part, then 3 example strings it should match and 2 it should correctly reject.
Constraints: Mentally trace your reject examples against the pattern before presenting them — a reject example the pattern would actually match defeats the point.

Image & Design Prompts (6)

Ready-to-paste prompts for Midjourney, DALL-E, Ideogram, or any text-to-image tool. These describe the scene and style rather than giving an instruction — that’s how image models actually read a prompt.

Two habits make these more reliable: naming what should not appear (clutter, watermark, extra limbs, text) as precisely as what should, and knowing where the --ar aspect-ratio flag works. It works in Midjourney and Ideogram, but not DALL-E/ChatGPT — there, just describe the shape in words (‘square format’, ‘wide landscape format’).

1. Studio Product Photography Style

A clean, e-commerce-ready product shot prompt for Midjourney, DALL-E, or Ideogram.

[PRODUCT] centered on a seamless white studio background, professional product photography, soft diffused lighting from the top left, subtle soft shadow beneath the product, high micro-detail, shot on an 85mm lens, commercial e-commerce style, no clutter, no reflections, no visible text or watermark –ar 1:1

Best for: Not sure which image generator to use? Compare options in Best AI Image Generators 2026.

2. Minimalist Logo Concept

A logo-style prompt tuned for flat, vector-friendly output rather than a photorealistic image.

Minimalist flat vector logo for [BRAND/COMPANY NAME], representing [CONCEPT, e.g. speed, growth, trust], simple geometric shapes, two-color palette ([COLOR 1] and [COLOR 2]), clean lines, icon-only, scalable design, no gradients, no drop shadows, no photorealistic texture, no extra text or tagline, white background –ar 1:1

3. Concept Art Character Design

A structured character-art prompt covering pose, style, and mood in one pass.

Concept art of [CHARACTER DESCRIPTION], [ART STYLE — e.g. digital painting, anime, fantasy realism], dynamic pose, detailed costume design, dramatic rim lighting, [MOOD/ATMOSPHERE], highly detailed, clean anatomy with correct hand and limb count, no watermark, no signature text –ar 2:3

4. Realistic Portrait Style

For a photorealistic portrait prompt with camera-specific detail that actually affects the output.

Photorealistic portrait of [SUBJECT DESCRIPTION], natural window lighting, shallow depth of field, shot on a 50mm lens, neutral background, candid expression, natural visible skin texture (not over-smoothed), correct hand anatomy if hands are visible, color graded like a professional headshot, no artifacts, no extra fingers –ar 3:4

5. Isometric App Icon

A tight prompt for UI/app-icon style output, which needs different keywords than a full scene.

Isometric 3D icon of [OBJECT/CONCEPT], soft pastel color palette, clean rounded edges, subtle drop shadow, centered composition, app icon style, no clutter, no baked-in text, white background, high resolution –ar 1:1

6. Cinematic Landscape

A wide-scene prompt with lighting and lens language for a more filmic result.

Cinematic wide-angle landscape of [SCENE DESCRIPTION], golden hour lighting, dramatic clouds, volumetric light rays, ultra-detailed, shot on an anamorphic lens, color graded like a film still, no visible people unless stated, no watermark, natural (not oversaturated) color grade –ar 16:9

Career & Resume Prompts (6)

Prompts for job applications, interview prep, and career decisions — written to need real input from you, not to fabricate achievements.

1. Resume Bullet Point Rewriter

Turns a vague duty-based bullet into an achievement with a number attached.

Role: You are a resume writer who turns duties into achievements without inventing numbers that aren’t there.
Task: Rewrite this resume bullet point to focus on a measurable achievement rather than a listed duty.
Format: One rewritten bullet, starting with a strong action verb, including a number or result only if it’s implied by the original.
Constraints: Don’t invent a statistic that isn’t in the original. If no result is implied at all, ask me for the actual outcome instead of guessing one.

Original: [PASTE BULLET POINT]

2. Cover Letter First Draft

A genuine-sounding first draft, not a generic template with the company name swapped in.

Role: You are a career coach who writes cover letters that sound like a specific person applying to a specific job, not a template.
Task: Write a cover letter for the [JOB TITLE] role at [COMPANY], using this background: [PASTE RELEVANT EXPERIENCE], and referencing this job description: [PASTE JOB DESCRIPTION].
Format: Under 300 words. Reference one specific detail from the job description or company. End with a clear next step.
Constraints: Confident but not over-the-top. Don’t claim experience or skills that weren’t in what I gave you.

3. Interview Answer Builder (STAR Method)

Structures a rambling story into a tight, interview-ready answer.

Role: You are an interview coach who structures real stories instead of writing generic-sounding answers.
Task: Help me structure an answer to: ‘[INTERVIEW QUESTION]’, using the STAR method (Situation, Task, Action, Result), based on this rough story: [PASTE YOUR EXPERIENCE/STORY].
Format: A tight answer, roughly 90 seconds when spoken aloud, in the S-T-A-R order.
Constraints: Use only details from the story I gave you. If the Result isn’t clear from my story, ask me what actually happened instead of inventing an outcome.

4. LinkedIn Headline & About Section

A headline and About section that reads like a person, not a keyword dump.

Role: You are a personal branding writer who avoids buzzword soup.
Task: Write a LinkedIn headline and About section for someone who is [ROLE/EXPERIENCE SUMMARY], looking for [GOAL — e.g. new roles in X, freelance clients, a career pivot into Y].
Format: Headline under 220 characters. About section under 200 words, written in first person.
Constraints: No generic phrases like ‘passionate professional’ or ‘results-driven’ — be specific about what I actually do, using details from what I gave you.

5. Salary Negotiation Script

A talking-points script for a specific negotiation conversation, not generic advice.

Role: You are a negotiation coach preparing me for one specific real conversation, not giving generic advice.
Task: My situation: current/offered salary is [AMOUNT], I’m asking for [TARGET AMOUNT], and my reasoning is [YOUR REASONS — market rate, competing offer, added scope, etc].
Format: 3-4 talking points I can actually say out loud, plus one polite response if they say no immediately.
Constraints: Base the talking points only on the reasoning I gave you — don’t invent a justification (like a competing offer) I didn’t mention.

6. Career Pivot Roadmap

A realistic step-by-step plan for switching fields, not just generic encouragement.

Role: You are a career strategist who gives realistic timelines, not motivational fluff.
Task: I currently work in [CURRENT FIELD/ROLE] and want to move into [TARGET FIELD/ROLE].
Format: A 6-month roadmap: the skills/certifications that actually matter for this specific switch, a realistic entry point given my background, and one concrete way to build a portfolio or proof of skill before applying.
Constraints: If my background makes a 6-month timeline unrealistic for this switch, say so directly instead of giving false encouragement.

Best for: For a deeper look at AI-related career paths, see AI Careers in 2026.

Study & Learning Prompts (6)

Prompts for learning something new, faster — built around active recall and real study structure, not just asking for a definition.

1. Explain a Concept Like I’m a Beginner

A first-principles explanation before you dive into denser material.

Role: You are a teacher who checks for understanding instead of just lecturing.
Task: Explain [CONCEPT/TOPIC] assuming I have no prior background.
Format: One real-world analogy first, then the actual technical definition. Under 200 words. End with one simple question I should be able to answer if I understood.
Constraints: If the analogy would oversimplify to the point of being misleading, note the one place it breaks down.

2. Create a Study Plan

Turns ‘I need to learn X’ into an actual week-by-week plan.

Role: You are a learning coach who builds plans with checkable milestones, not just a topic list.
Task: Create a [NUMBER]-week study plan to go from [CURRENT LEVEL — e.g. complete beginner] to [GOAL LEVEL — e.g. able to pass X exam / build a basic project] in [SUBJECT/SKILL].
Format: Weekly milestones, each with a specific, checkable outcome (something I can verify I did, not just ‘study chapter 3’).
Constraints: If [GOAL LEVEL] is unrealistic in [NUMBER] weeks given a true beginner starting point, say so and suggest a more realistic timeline instead.

Best for: If the subject is AI itself, see How to Learn AI in 2026 for a ready-made path.

3. Quiz Me on This Topic

Active recall instead of passive re-reading.

Role: You are a study partner running an active-recall quiz, not just handing me answers.
Task: Quiz me on [TOPIC] with 8 questions, mixing multiple choice and short answer.
Format: Ask one question at a time and wait for my answer before revealing whether I’m right and explaining the correct answer briefly.
Constraints: Don’t move to the next question until I’ve responded to the current one.

4. Summarize a Long Document

Condenses a long text into something you can actually retain.

Role: You are an editor who condenses without distorting.
Task: Summarize the text below.
Format: (1) a 2-sentence overview, (2) the 5 most important points as a bulleted list, (3) one thing a reader is likely to misunderstand or overlook.
Constraints: Summarize only what’s in the text — don’t add outside information, context, or opinion that isn’t stated.

Text: [PASTE TEXT]

5. Flashcard Generator

Turns notes into spaced-repetition-ready flashcards in one pass.

Role: You are building flashcards for spaced repetition, where each card must test exactly one fact.
Task: Turn the notes below into 15 flashcards.
Format: Question / Answer pairs, one per line, suitable for an app like Anki. Each answer should be one clear fact — no multi-part answers.
Constraints: If a note bundles multiple facts together, split it into separate cards rather than cramming them into one answer.

Notes: [PASTE NOTES]

6. Compare Two Concepts

A side-by-side breakdown for anything that’s easy to confuse.

Role: You are a teacher explaining the exact point where two similar ideas diverge.
Task: Explain the difference between [CONCEPT A] and [CONCEPT B].
Format: A comparison table covering definition, one real-world example of each, and the single biggest point of confusion between the two.
Constraints: Name the actual root cause of the confusion — not just ‘they sound similar,’ but what specifically overlaps.

Productivity & Planning Prompts (6)

Prompts for turning a messy to-do list, meeting, or inbox into something you can actually act on.

1. Daily Priority Planner

Turns a messy to-do list into an actual ordered plan for the day.

Role: You are a productivity coach applying a real framework, not just reordering my list randomly.
Task: Here’s my to-do list for today: [PASTE TASKS]. Prioritize it using the Eisenhower method (urgent/important).
Format: A suggested order to tackle tasks in, with each one tagged by its quadrant. Flag anything that looks like it could be delegated, delayed, or dropped.
Constraints: If two tasks both look urgent-and-important, tell me which one to do first and why, rather than leaving the tie unresolved.

2. Meeting Notes to Action Items

Extracts decisions and owners from a messy transcript or notes dump.

Role: You are someone who takes minutes that people actually follow up on.
Task: Turn these meeting notes into a clean summary.
Format: Three sections — Key Decisions Made, Action Items (with an owner and rough deadline if mentioned), and Open Questions still needing answers.
Constraints: Don’t invent an owner or deadline that wasn’t actually stated — write ‘unassigned’ or ‘no deadline given’ instead of guessing.

Notes: [PASTE NOTES/TRANSCRIPT]

3. Decision-Making Framework

Structures a hard decision instead of going in circles about it.

Role: You are a decision analyst who structures hard choices instead of giving a one-line opinion.
Task: I’m trying to decide between [OPTION A] and [OPTION B] for [DECISION CONTEXT].
Format: First, name the 2-3 criteria that actually matter here. Then a pros/cons table for each option against those criteria. Then state which option the analysis points toward.
Constraints: End with the one specific condition under which the other option would actually be the better call — a recommendation with no caveat is usually hiding uncertainty, not resolving it.

4. Weekly Reflection Prompt

A structured end-of-week review instead of a vague journaling prompt.

Role: You are running a structured weekly review, not a vague journaling prompt.
Task: Ask me 5 weekly reflection questions: what went well, what didn’t go as planned, one lesson learned, one thing to change next week, and one thing I’m looking forward to.
Format: Ask them one at a time and wait for my answer before asking the next.
Constraints: After all 5 answers, give me one short observation connecting two of my answers, if a real connection exists — don’t force one if there isn’t.

5. Automation Workflow Idea Generator

Surfaces where a repetitive task could actually be automated, not just AI-assisted.

Role: You are an automation consultant who names real tools, not vague ‘you could automate this’ advice.
Task: I do this task repeatedly: [DESCRIBE THE REPETITIVE TASK]. Suggest 3 ways to automate or streamline it.
Format: Ranked from ‘quick fix, no-code’ to ‘more involved but higher-leverage,’ each naming a specific real tool or approach.
Constraints: If the task genuinely can’t be automated well (too much judgment involved), say that instead of forcing a fake automation idea.

Best for: For a broader look at what AI agents can actually automate, see AI Agents & Automation: What They Are.

6. Email Inbox Triage

Turns a wall of unread emails into a sorted action list.

Role: You are triaging my inbox the way a sharp executive assistant would.
Task: Here are the subject lines and first lines of my unread emails: [PASTE LIST]. Sort them.
Format: Three buckets — Needs a reply today, Can wait until this week, No action needed / can archive. Flag anything time-sensitive or from an important sender.
Constraints: If a subject line alone is ambiguous, default to the more cautious bucket — don’t archive something that might be time-sensitive.

Research & Analysis Prompts (6)

Prompts for thinking something through properly — built to surface disagreement and weak points, not just validate whatever you already believe.

1. Pros and Cons Analysis

A genuinely balanced breakdown instead of a one-sided argument.

Role: You are an analyst whose job is to stress-test an idea, not sell it to me.
Task: Give me a balanced pros and cons analysis of [DECISION/TOPIC].
Format: At least 4 points on each side. Then state what would need to be true for the pros to clearly outweigh the cons.
Constraints: Every con must be a genuine drawback, not a disguised positive (‘it’s so good people might get addicted to it’). If you can’t find 4 genuine cons, say so rather than padding the list.

2. Fact-Check This Claim

Pushes back on unverified claims instead of confirming them by default.

Role: You are a skeptical fact-checker, not a validator.
Task: I came across this claim: ‘[PASTE CLAIM]’. Assess how plausible it is.
Format: A plausibility read based on what’s generally known, what kind of source would actually be needed to verify it properly, and a clear statement of your confidence level.
Constraints: If you’re not confident either way, say that plainly instead of leaning toward agreement by default — agreeing by default is a known failure mode, not a safe one.

3. Summarize Multiple Sources on a Topic

Synthesizes several inputs into one coherent view instead of reading them one by one.

Role: You are synthesizing sources the way a research analyst would, not just concatenating summaries.
Task: I’m giving you information from multiple sources on [TOPIC].
Format: First, the points of general agreement across all sources. Then, explicitly call out any point where the sources actually disagree or contradict each other.
Constraints: Don’t smooth over a real disagreement to make the sources sound more aligned than they are.

Source 1: [PASTE]
Source 2: [PASTE]
Source 3: [PASTE]

4. Data Interpretation Helper

Turns a table of numbers into a plain-language read on what’s actually going on.

Role: You are a data analyst who separates real signal from noise.
Task: Here’s a data table: [PASTE DATA]. Identify the most notable patterns.
Format: The 3 most notable patterns or outliers, a plausible explanation for each, and which ones might just be noise or a small sample size rather than a real trend.
Constraints: Reason through sample size and possible confounders before concluding something is a real trend, not just a coincidence.

5. Trend Analysis

A structured read on where something is heading, not just a description of where it is now.

Role: You are a strategist forecasting a trend, not just describing the present.
Task: Analyze the current trend in [INDUSTRY/TOPIC].
Format: What’s changed in the last 12 months, the most likely driver behind that change, one prediction for the next 6-12 months, and one factor that could completely change that prediction if it shifts.
Constraints: State your prediction’s confidence level honestly — a forecast presented with total certainty is usually overconfident.

6. Devil’s Advocate Review

Stress-tests an idea by deliberately arguing against it before you commit.

Role: You are the smartest, most well-informed critic of this plan — not a strawman.
Task: I’m planning to do this: [PLAN/IDEA]. Argue against it as strongly as a genuine skeptic would.
Format: The single strongest argument against the plan, then a note on which part of that argument is hardest to dismiss.
Constraints: No strawmanning — if you can’t find a genuinely strong objection, say the plan is solid rather than manufacturing a weak one.

How to Get More Out of Any Prompt

A prompt from a list is a starting point, not a finished tool. If you want to adapt one of these — or write your own from scratch — keep the same four parts in mind: a specific Role and Task, an exact Format, and Constraints that say what not to do.

Then don’t accept the first answer if it’s generic. Ask the AI to be more specific, cut the fluff, or try a different angle — the best results usually come from a short back-and-forth, not a single message.

Frequently Asked Questions

Do these prompts work with ChatGPT, Claude, and Gemini equally well?

Most of the text prompts (writing, business, coding, career, study, productivity, research) work well across ChatGPT, Claude, and Gemini with little to no change — the underlying instruction matters more than the specific model. The image prompts are the exception: Midjourney, DALL-E, and Ideogram each interpret style keywords slightly differently, so treat those as a starting point and adjust based on the tool you’re using.

Why do some prompts have brackets like [TOPIC] in them?

Brackets mark the part you need to fill in with your own specifics. A prompt like ‘Explain [CONCEPT]’ only works once you replace [CONCEPT] with an actual topic — copying it exactly as written with the brackets left in will confuse the AI, not help it.

Why do the prompts say “Role / Task / Format / Constraints” instead of just one instruction?

Because separating those four parts consistently produces better output than one run-on sentence — the model doesn’t have to infer your intent, it’s told directly.

It also makes edits easy: change the Role to shift the tone, change the Format to change the output structure, and add to Constraints to fix a recurring problem — like the AI adding information you didn’t give it.

Why isn’t my prompt working the way I expected?

The most common cause is a vague or missing detail the AI has to guess at — audience, tone, length, or format. Add the specific constraint you actually care about (word count, tone, format) directly into the prompt rather than assuming the AI will pick the right default.

How often is this page updated?

As AI models change, some prompt phrasing stops working as well and needs a refresh — this page is reviewed periodically rather than being a static, one-time list.

Last updated: September 2026 — rewritten using the Role/Task/Format/Constraints structure for more consistent output, with one-click copy buttons added to every prompt. Explore our full AI Tools directory to find the right tool for any of the prompts above.

Curious how those Copy Prompt buttons were built — and the two silent WordPress bugs we had to fix along the way? Read the full debugging postmortem.