Most lists of AI prompt examples have the same problem: they show you the prompt but not why it works. You copy it, get a mediocre result, and wonder what went wrong. This guide is different. Every section shows a weak version of the prompt alongside a strong one, and explains the specific change that […]
Most lists of AI prompt examples have the same problem: they show you the prompt but not why it works. You copy it, get a mediocre result, and wonder what went wrong.
This guide is different. Every section shows a weak version of the prompt alongside a strong one, and explains the specific change that makes the difference. By the end, you won’t just have prompts to copy — you’ll understand what separates a prompt that gets a generic response from one that gets something genuinely useful.
Before the examples: every strong prompt does some combination of five things. It sets a role (“act as a senior copywriter”), states a clear task (“write a subject line”), provides context (“for a SaaS product launching a free trial”), defines a format (“under 10 words”), and applies constraints (“no exclamation marks, no urgency bait”).
You rarely need all five. But the more of them you include, the less the AI has to guess — and guessing is where generic outputs come from.
Weak: Write a blog post intro about remote work.
Strong: Write a 3-sentence blog intro for an article titled “Why Remote Work Feels Harder Than It Should.” The audience is mid-level managers at tech companies who are skeptical of remote work but open to evidence. Open with a surprising statistic or counterintuitive observation — no motivational fluff.
Why it works: The strong version specifies who it’s for, what emotional tone to aim for, and what to avoid — cutting out a full paragraph of revision.
Weak: Rewrite this paragraph to sound better.
Strong: Rewrite this paragraph to be 30% shorter without losing any information. Remove filler phrases, passive voice, and anything that doesn’t serve the reader. Keep the original meaning exactly: [paste paragraph]
Why it works: “Sound better” is unmeasurable. A word count target and specific rules give the AI something to optimize against.
More writing examples:
Generate 10 headline options for an article about [topic]. Include: 2 how-to headlines, 2 numbered list headlines, 2 question headlines, 2 contrarian takes, and 2 “what no one tells you” angles. Audience: [describe].
I’m writing about [topic]. Give me 5 counterintuitive angles that most writers miss — facts, stats, or perspectives that would surprise someone who thinks they already know this topic.
Summarize this article in 3 sentences. First sentence: the main claim. Second: the strongest supporting evidence. Third: the practical takeaway. [paste article]
Act as a developmental editor. Read this draft and tell me: (1) where the argument is weakest, (2) what a skeptical reader would ask for that’s missing, (3) which section I should cut entirely. Don’t rewrite anything — just diagnose: [paste draft]
Weak: Write a marketing email for our product launch.
Strong: Act as a direct-response email copywriter. Write a launch email (under 200 words) for [Product Name], a [one-line description], aimed at [audience]. The reader’s main hesitation is [objection]. Lead with a short story or surprising fact, address the hesitation mid-email, and close with one CTA: [CTA text]. No subject line — just the body.
Why it works: Naming the reader’s hesitation forces the email to actually sell rather than announce. Most launch emails forget this.
Weak: Write 5 subject lines for a promotional email.
Strong: Write 10 email subject lines for a promotional email offering [offer] to [audience]. Include: 3 curiosity-gap lines, 2 direct benefit lines, 2 social proof lines, 2 urgency lines (no fake deadlines), and 1 weird/unexpected line. Under 50 characters each. No exclamation marks.
More marketing examples:
Analyze this landing page copy and tell me: what’s the main value proposition, what’s missing that a skeptical visitor would want to know, and what single change would most improve conversion. [paste copy]
Write 3 versions of a Google ad for [product/service]: one focused on price, one on speed, one on trust. Each: 1 headline (30 chars max) + 1 description (90 chars max).
Act as a cold email specialist. Write a 5-sentence cold email to [job title] at [company type] about [offer]. Rule: no opener about being “excited” or “reaching out.” Get to the point in the first sentence.
I have a product that [does X]. Write 5 Instagram captions: one educational, one behind-the-scenes, one testimonial-style, one with a question to drive comments, one with a strong opinion. Max 150 words each.
Weak: Summarize this meeting transcript.
Strong: Summarize this meeting transcript into three sections: (1) Decisions made — list each as a one-liner, (2) Action items — who owns what and by when, (3) Open questions — anything unresolved needing follow-up. Use bullet points. Skip small talk. [paste transcript]
Why it works: A structured output format means the summary is immediately useful — not a wall of prose someone has to re-read to extract the information.
More productivity examples:
Help me write a professional reply to this email. The sender is [their role] and they are [frustrated / confused / asking for something unreasonable]. I want to [your goal]. Keep it under 100 words, firm but polite, end with a clear next step. Here’s their email: [paste]
Turn this messy list of tasks into a prioritized action plan. Flag anything urgent, anything that can be delegated, and anything that can wait until next week. Tasks: [paste list]
Draft a project brief for [project name]. Include: objective, key deliverables, timeline, stakeholders, and the one metric we’ll use to measure success. Keep it to one page. Here’s what I know so far: [paste notes]
I have 6 hours today. Here are my tasks: [list]. Create a time-blocked schedule that prioritizes deep work in the morning, accounts for a 30-minute lunch, and leaves a 30-minute buffer at the end for overruns.
Weak: Fix my code.
Strong: I’m getting this error: [paste error]. Here’s the relevant code: [paste code]. I’m using [language/framework/version]. Explain what’s causing it in plain English, then give me the corrected code with comments on what you changed and why.
Why it works: The error message, code snippet, and environment give the AI what it needs to actually diagnose the problem rather than guess.
More coding examples:
Review this function for bugs, edge cases, and performance issues. Rewrite it with improvements and comment each change. Language: [language]. Function: [paste code]
Write a [language] function that [does X]. Requirements: handles [edge case], returns [format], includes error handling for [scenario]. Add inline comments. No external libraries unless specified.
Explain this code to a non-technical stakeholder in 3 sentences. What does it do, why does it matter, and what would happen if it broke? [paste code]
Weak: Research the electric vehicle market.
Strong: Give me a structured briefing on the electric vehicle market in Europe as of 2026. Include: (1) market size and growth rate, (2) top 3 players and their market share, (3) biggest consumer barrier to adoption, (4) one trend most analysts are underestimating. Use plain language, cite specific numbers where possible.
Why it works: A “briefing” with specific sub-questions produces something actionable. “Research X” produces an essay.
More research examples:
Compare [Option A] and [Option B] across: cost, implementation complexity, scalability, and vendor risk. Output as a table. Add a one-paragraph recommendation for a company that is [describe company type].
I’m preparing for a meeting with [company/person]. Give me: (1) 5 things I should know before walking in, (2) 2 things that might come up that I should have an answer for, (3) 1 question I should ask that most people wouldn’t think to.
Steelman this argument — give me the strongest possible version of it, as if made by someone who deeply believes it: [paste argument]
I’m writing about [topic]. What are the most commonly cited statistics in this space? Flag any that are frequently misquoted, outdated, or have methodological problems worth knowing about.
Weak: Explain machine learning to me.
Strong: Explain machine learning to me like I’m a marketing manager who understands data but has no engineering background. Use one real-world analogy, give one concrete example of how a company uses it today, and end with the one thing I should actually understand vs. the things I can safely ignore.
Why it works: The audience definition and the “what to ignore” constraint force a genuinely useful explanation rather than an encyclopedic one.
More learning examples:
I want to learn [skill] from zero. I have [X hours] per week and [timeframe]. Give me a realistic learning plan: what to start with, the order to tackle things, what resources to use at each stage, and the first thing I should do today.
Quiz me on [topic]. Ask one question at a time. After I answer, tell me if I’m right or wrong, explain the correct answer briefly, then ask the next one. 10 questions, increasing in difficulty.
Give me the 20% of knowledge about [topic] that covers 80% of what matters for someone in [my role]. Skip academic depth — I need to be competent and credible, not an expert.
Act as a thoughtful friend who happens to have expertise in [career/finance/relationships]. I want honest advice, not validation. Here’s my situation: [describe]. What am I missing, and what would you actually do in my position?
Help me plan a [trip/event/project]. Constraints: [budget, dates, preferences]. Give me a day-by-day itinerary / checklist / timeline with specific recommendations — not general tips.
I need to write a [toast / speech / tribute] for [occasion and person]. Here are 5 things that capture who they are: [list]. Write something warm, specific, and gets a genuine laugh somewhere in the middle. Under 3 minutes to read aloud.
Help me write a LinkedIn post about [topic or experience]. Honest and direct — not “thought leadership” or humble-brag. Under 150 words. End with a question to get comments.
Look at every strong prompt above and you’ll see the same thing: they don’t ask the AI to figure out what you want. They tell it the audience, the format, the constraints, and often what to avoid. The more of that work you do upfront, the less editing you do afterward.
That’s the whole game.
If you want a prompt built specifically for your situation — without filling in a template — our free AI Prompt Generator does it from a plain-English description in seconds.
And if you want to go deeper on the structure behind every strong prompt above, the guide on how to write the perfect AI prompt breaks down the five-part framework with more examples.
Need prompts for a specific workflow? The WePrompt Prompt Directory has 50+ tested, ready-to-use prompts across writing, marketing, research, job hunting, and more.
Save yourself the work
500+ tested prompts, ready to copy.
Instead of crafting the perfect prompt from scratch, browse the directory, copy what you need, and customise it in seconds.
Browse the Prompt Directory →
Leave a reply
Your email address will not be published. Required fields are marked *