Practical prompts,
poured by category.

“Act as my AI research assistant. Summarize this topic in simple language, give me 5 key points, and suggest 3 content ideas for Instagram, LinkedIn, and a newsletter.”

“You are an Instagram content strategist. Take this single idea and turn it into a 7-slide carousel outline. For each slide, give me a punchy hook, a one-line body, and a visual suggestion. Keep the tone calm, smart, and useful.”

“Write a 900-word blog post on [TOPIC] in a warm, editorial, newspaper-style tone. Open with a small story. Use short paragraphs, clear subheadings, and end with a one-line takeaway. Avoid hype, jargon, and marketing language.”

“Here are my notes from the week: [NOTES]. Turn them into a 5-section weekly newsletter with: one biggest story, one tool worth trying, one prompt, one skill, and one short takeaway. Keep each section under 80 words.”

“Summarise the following meeting transcript into: (1) decisions made, (2) action items with owners, (3) open questions, and (4) a one-paragraph summary I can paste into Slack. Be concise. Do not invent details.”

“Build me a 4-week learning plan to get genuinely good at [SKILL]. Assume I have 30 minutes a day. For each week, give me 3 things to read or watch, 1 small project, and 1 honest self-check question.”

“Act as a brand strategist. Using publicly available information, give me a one-page snapshot of [COMPANY]: positioning, target audience, top three products, pricing model, and one weakness a smaller competitor could exploit. Be specific and avoid fluff.”

“Rewrite this email so it stays clear and direct but removes any defensive or tense tone. Keep it short. Keep the meaning exactly the same. Do not add new information.”

“Here is my current weekly workflow: [DESCRIBE]. Identify the 3 steps most worth automating with AI today. For each, tell me the tool I should use, the rough setup cost in hours, and the time it will save per week.”

“Take this blog post and turn it into an 8-tweet thread. Start with a strong hook. Use one idea per tweet. End with a soft CTA. Keep it conversational. no all-caps, no emojis, no clickbait.”

“Before I write on [TOPIC], help me plan the research. Give me: (1) the 5 most credible sources, (2) the 3 most contested claims, (3) the 3 strongest counterarguments, and (4) 5 angles I should consider before forming an opinion.”

“Write 3 short ad copy variations for [PRODUCT] aimed at [AUDIENCE]. Tone: calm, intelligent, helpful. not loud or pushy. Each variation should be under 30 words. Lead with a benefit, not a product feature.”

“Act as a content strategist. Turn this rough idea into a practical content brief: [IDEA]. Include the target audience, core problem, main takeaway, 5 key points to cover, suggested structure, and one strong opening angle. Keep it clear and specific.”

“Act as a data analyst. Review the following data: [DATA]. Identify the 5 most important patterns, anomalies, or changes. Explain what each could mean in simple language, then give me 3 practical questions I should investigate next. Do not invent missing data.”

“Here are my current tasks: [TASKS]. Rank them by impact, urgency, and effort. Give me the top 3 tasks to do first, explain why each matters, and suggest what I can postpone, delegate, or remove. Keep the recommendation practical.”

“Act as a customer research strategist. Using this business or product information: [DETAILS], create a practical customer persona covering their goals, problems, buying triggers, objections, common questions, and what would make them trust the product. Clearly separate assumptions from known information.”

“Take this core idea: [IDEA]. Turn it into: 1 LinkedIn post, 1 Instagram caption, 1 short X post, 1 newsletter section, and 3 short video ideas. Keep the central message consistent while adapting the format and tone for each platform.”

“Turn this single idea into a complete content pack I can actually publish from: [IDEA]. Audience: [AUDIENCE]. Goal: [GOAL]. Give me: (1) a working title and one-line promise, (2) a blog outline with 5 H2s and 2 bullet points under each, (3) a 120-word email that can stand alone, (4) 5 opening hooks, (5) 3 clear CTAs, and (6) 3 things I should leave out so the pack stays focused. Keep the tone practical and plain. Do not pad with extra formats I did not ask for.”

“Research this topic using only the material I provide. Topic: [TOPIC]. Sources: [SOURCES]. If a source is missing, work only with what is here. Do the following: (1) summarise what the sources actually say, (2) list 5 claims that are clearly supported, with a short quote or paraphrase for each, (3) list anything that is inferred rather than stated, (4) list open questions and missing facts, and (5) give me 3 honest next research steps. Never invent names, dates, numbers, or studies. If you are unsure, write "unknown" instead of guessing.”

“Here are my messy notes: [NOTES]. Turn them into a clear action plan. Do not add tasks I did not mention. Give me: (1) the real outcome these notes are pointing to, in one sentence, (2) a numbered plan in the order I should do the work, (3) the single next action I can do in the next 25 minutes, (4) any decisions still blocking progress, and (5) what can wait until later. Keep each item short and specific. If a note is unclear, ask one clarifying question instead of filling the gap.”

“Help me build a weekly AI workflow for my business. Business type: [BUSINESS]. The work I repeat most: [WORKFLOW]. Tools I already use: [TOOLS]. Time I can spare: [TIME] per week. Design a 5-day plan with one AI-assisted task per day. For each day, give me: the task, the input I should paste in, the prompt I should run, the output I should keep, and a 20-minute time box. End with one weekly review question so I can tell if the workflow is actually saving time. Keep it simple enough to reuse next week. Do not recommend a stack of new tools.”

“Here are 2 or 3 writing samples that sound like me: [SAMPLES]. Here is the AI-written draft I want rewritten: [DRAFT]. First, extract 6 voice rules from my samples: sentence length, words I use, words I avoid, how I open, how I land a point, and how formal I am. Then rewrite the draft so it follows those rules. Keep my meaning. Cut hype, filler, and generic AI phrasing. After the rewrite, list the 5 biggest changes you made. If a sample is too short to judge a rule, say so instead of inventing a style.”

“Turn this AI conversation into a reusable system I can use again: [CHAT]. Extract: (1) the goal we were trying to achieve, (2) important context and constraints, (3) decisions we made and why, (4) rules or preferences that should be preserved, (5) prompts or instructions worth reusing, and (6) a short step-by-step process for repeating the work. Separate confirmed decisions from suggestions. Do not invent missing context. If something is unclear, mark it as unclear.”

“Stress-test this AI-generated answer before I rely on it: [ANSWER]. Check it for: (1) claims that need evidence, (2) assumptions presented as facts, (3) missing context that could change the conclusion, (4) contradictions or logical gaps, (5) outdated or uncertain information, and (6) important points it failed to address. Then give me a corrected version only where necessary. Clearly label anything you cannot verify instead of guessing.”

“I have finished this project: [PROJECT]. Here is what I actually did: [WORK_NOTES]. Turn it into a repeatable process for the next time I do similar work. Give me: (1) the goal, (2) the steps in the order they should happen, (3) inputs needed at each important step, (4) decisions or checks that should not be skipped, (5) common mistakes revealed by this project, and (6) a short reusable checklist. Base it only on what I actually did. Do not add imaginary steps just to make the process look complete.”

“I have two AI-generated outputs for the same task. Output A: [OUTPUT_A]. Output B: [OUTPUT_B]. My actual goal is: [GOAL]. Compare them on accuracy, usefulness, clarity, completeness, and fit for my goal. Identify what each does better and what should be discarded. Then create a third version that combines only the strongest parts of both. Do not average the answers or preserve a weak point just because it appears in both. If either output contains an unsupported claim, flag it rather than carrying it forward.”

“Help me prepare for this difficult conversation: [SITUATION]. The person I am speaking with is: [PERSON/ROLE]. What I want to achieve: [OUTCOME]. What I am worried they may say: [CONCERNS]. Prepare: (1) the three points I most need to communicate, (2) a calm opening, (3) likely objections and how I can respond, (4) questions I should ask, (5) things I should avoid saying, and (6) a concise closing. Keep the language natural and respectful. Do not assume the other person is hostile or agree with me automatically.”

“Use the uploaded photo as the identity reference and recreate it as if the photograph had genuinely been taken in the 1980s. Keep the person's identity recognisable: preserve their facial structure, skin tone, age, body proportions, and natural features. Do not replace their face with a generic 1980s-looking person. Change the styling to feel authentically 1980s: period-appropriate hairstyle, clothing, accessories, background, lighting, and photography style. Use realistic analogue film characteristics such as subtle grain, slightly faded colours, natural film softness, and the look of a real photograph from the decade. Choose details that fit the person naturally rather than piling on every stereotypical 80s element. The final image should look like a genuine 1980s photograph of this same person, not a modern photo with an 80s filter. Avoid modern clothing, modern lighting, contemporary backgrounds, excessive digital sharpness, plastic-looking skin, exaggerated retro effects, text, logos, or artificial-looking facial changes.”

“Turn this AI news, announcement, or release into a useful brief: [SOURCE]. If the source is incomplete, work only with what is here. Give me: (1) what happened, in one or two plain sentences, (2) what actually changed for users, products, or workflows, (3) why it matters, (4) who it matters to, and who can ignore it for now, (5) confirmed facts from the source, (6) claims, marketing language, or details that still need verification, (7) what a normal user can actually do with the change, and (8) a short plain-English takeaway. Do not invent missing details. Do not treat company claims as independently verified facts. If something is unclear, label it as unclear.”

“Help me test an AI model on my real work, not a generic demo. Task: [TASK]. Sample input: [INPUT]. What a good output looks like: [SUCCESS]. Constraints: [CONSTRAINTS]. Previous output for comparison, if I have one: [PREVIOUS_OUTPUT]. Model I want to test: [MODEL]. Design a simple repeatable test I can run again later. Give me: (1) the exact prompt or steps I should use, (2) 5 evaluation criteria that matter for this task, (3) a short scorecard template, (4) how to compare the new result with my previous output if one exists, and (5) what the result would mean in practical terms for my workflow. Judge usefulness for this specific job. Do not declare a universally best model.”

“Help me turn this AI agent idea into a safe, clearly bounded job: [IDEA]. Context: [CONTEXT]. Tools or systems involved: [TOOLS]. Turn it into a practical starting specification with: (1) the exact job in one sentence, (2) what the agent is allowed to do, (3) what it must never do, (4) what information or data it can access, (5) which actions require human approval, (6) what counts as success, (7) what should happen when something goes wrong, and (8) when a human should take over. Keep the scope small enough to test. Do not present the agent as fully autonomous or infallible.”

“Here is a messy voice-note transcript: [TRANSCRIPT]. Audience: [AUDIENCE]. Intended format: [FORMAT]. Turn it into useful content without removing my natural way of speaking. Give me: (1) the main idea in one sentence, (2) the useful points worth keeping, (3) missing context or unclear sections I should fix, (4) a clean publishable draft, (5) 3 possible hooks, and (6) a short social-media version. Keep my tone. Cut filler and repetition, but do not make it sound corporate, overly polished, or obviously AI-written. Do not invent personal experiences, opinions, facts, or details that are not in the transcript.”

“Help me turn this trending AI topic into original content, not another rewrite of the same announcement: [TOPIC]. Audience: [AUDIENCE]. What I already know or have seen: [NOTES]. Give me: (1) what everyone is already saying, (2) the most interesting unanswered question, (3) 3 less-obvious angles, (4) a practical angle for my audience, (5) one useful contrarian question without forcing a contrarian opinion, (6) one original content concept with a working title and one-line promise, and (7) the best format for it and who it is for. Prioritise a useful perspective, explanation, experiment, comparison, or practical takeaway. Do not invent facts. Do not pretend an unsupported angle is established.”

“I want to build or change something with an AI coding tool. Before writing any code, help me create a clear specification. Ask me only the questions necessary to understand: (1) what I want to build, (2) who will use it, (3) the expected behaviour, (4) inputs and outputs, (5) important edge cases, (6) technical constraints, and (7) what should explicitly NOT be changed. Then turn my answers into a concise implementation specification that an AI coding agent can follow. Do not write code yet.”

“I have a task I repeat regularly: [DESCRIBE THE TASK]. Help me turn it into a practical AI-assisted workflow. Break it into: (1) what I should provide to AI, (2) what AI should do, (3) what I should review or approve, (4) what can safely be automated, and (5) what should always remain under human control. Keep the workflow simple enough that I could actually use it repeatedly. If part of the process is not suitable for automation, say so instead of forcing AI into it.”

“Here is an AI-generated answer: [PASTE AI OUTPUT]. Improve it without changing the underlying meaning. Check for: unnecessary wording, unclear explanations, unsupported claims, repetition, missing context, and overly confident statements. Then give me: (1) the improved version, (2) the most important changes you made, and (3) anything that still needs human verification. Keep the final answer clear, natural, and practical.”

“I am planning to do this: [DESCRIBE MY PLAN]. Before I execute it, act as a critical reviewer. Identify: (1) assumptions I may be making, (2) obvious risks, (3) missing steps, (4) things that could fail, and (5) information I should verify first. Then suggest specific improvements to the plan. Do not reject the plan just for the sake of criticism. Focus on issues that could materially affect the outcome.”