50 AI Prompt Examples That Actually Work (2026)

Email and communication prompts (1-5)

**1. Professional email from rough notes.** "I need to email [name/role] about [topic]. Here are my rough notes: [paste notes]. Write a professional email that is concise, clear, and ends with a specific next step. Tone: [formal/friendly/direct]. Maximum 200 words." — This works because it provides context (recipient, topic), raw material (notes), format constraints (word limit), and tone guidance. The AI has everything it needs to produce a targeted draft.

**2. Difficult conversation email.** "I need to deliver [type of news — e.g., project delay, budget cut, role change] to [audience]. The key facts are: [list facts]. Write an email that leads with empathy, states the situation clearly without hedging, explains the reasoning, and offers a concrete next step. Do not use corporate jargon or passive voice." — Difficult communications are where AI assistance is most valuable. The constraint against jargon and passive voice prevents the most common failure mode: watered-down language that buries the message.

**3. Meeting follow-up.** "Here are my raw meeting notes: [paste notes]. Write a structured follow-up email with three sections: Key Decisions (bullet points), Action Items (who, what, by when), and Next Meeting agenda items. Use the attendees' first names. Keep it under 300 words." — Specifying the exact structure ensures the output is immediately usable. The "who, what, by when" format for action items prevents vague follow-ups.

**4. Cold outreach email.** "Write a cold outreach email to [target role] at [type of company]. My value proposition is: [one sentence]. Reference a specific challenge that [target role] typically faces: [challenge]. Keep it under 150 words. No fluff, no 'I hope this finds you well,' no fake personalisation. End with a low-friction ask (not 'schedule a 30-minute call')." — The explicit prohibitions ("no fluff," "no fake personalisation") steer the AI away from the generic patterns that make AI-generated outreach instantly recognisable.

**5. Internal announcement.** "Draft an internal announcement for [audience — e.g., all-hands, department, team] about [change/news]. Key details: [list]. Tone: [transparent, optimistic, measured]. Address the most likely question employees will have. Include a clear 'what this means for you' section and state where people can ask further questions." — Anticipating audience questions within the prompt produces dramatically more useful outputs than a simple "announce X." For more techniques like these, explore the [Prompt Template Library](/tools/prompt-library).

Research and analysis prompts (6-15)

**6. Structured research brief.** "Research [topic] and produce a structured brief with these sections: Executive Summary (3 sentences), Key Facts (10 bullet points with sources where possible), Arguments For [position], Arguments Against [position], Open Questions, and Recommended Next Steps. I need this for [context/purpose]." — Specifying the exact output structure and stating the purpose ensures relevant, actionable research rather than a generic overview.

**7. Competitive analysis.** "Analyse the competitive landscape for [product/service] in [market]. Compare the top 5 competitors across: pricing model, core features, target customer, key differentiator, and most common customer complaint. Format as a comparison table. Then provide 3 bullet points on underserved needs in this market." — The table format makes the output immediately useful for presentations. The "underserved needs" addendum turns a descriptive analysis into a strategic one.

**8. Document summarisation with opinion.** "Here is a [document type — report, article, paper]: [paste or upload]. Provide: 1) A 3-sentence summary, 2) The 5 most important claims with page/section references, 3) The strongest argument, 4) The weakest argument, 5) What questions this document does NOT answer. Be critical, not just descriptive." — The "be critical, not just descriptive" instruction is essential. Without it, AI summaries tend to be neutral restatements that add little analytical value.

**9. Data interpretation.** "Here is a dataset: [paste or describe data]. Identify the 3 most interesting patterns or anomalies. For each, explain what it might mean and what additional data you would want to investigate further. Assume I am presenting to [audience] — frame the findings at an appropriate level of technical detail." — Asking for "what additional data you would want" transforms the AI from a reporter into a thinking partner.

**10. Literature review synthesis.** "I have read these sources on [topic]: [list titles/summaries]. Synthesise the key themes across all sources. Identify where they agree, where they contradict each other, and what gaps exist. Organise by theme, not by source." — "Organise by theme, not by source" is the critical instruction. Without it, AI produces a source-by-source summary rather than a genuine synthesis.

**11-15: Additional research prompts.** "Compare [framework A] vs [framework B] for [specific use case]. Structure as: overview, strengths, weaknesses, best-fit scenarios, verdict." | "Extract all quantitative claims from [document] and assess the strength of evidence for each on a scale of strong/moderate/weak." | "I'm preparing for a meeting about [topic]. Generate 10 questions I should be prepared to answer, with suggested responses for each." | "Review this [proposal/plan] as if you were a sceptical board member. What are the 5 biggest risks or weaknesses, and what would you need to see to be convinced?" | "Analyse [trend/data] and generate three possible scenarios: optimistic, base case, and pessimistic. Include the assumptions behind each." These research prompts follow the [CONTEXT Framework](/context-framework) principles — providing circumstance, specifying objectives, and setting explicit expectations for the output format.

Content creation prompts (16-20)

**16. Blog post outline from a thesis.** "My thesis is: [one sentence argument]. My audience is: [description]. Create a blog post outline with: a hook opening (not a question), 5-7 subheadings that build the argument progressively, key points under each subheading, a conclusion that ties back to the opening. Target length: [word count]. Tone: [authoritative/conversational/practical]." — Starting from a thesis rather than a topic produces focused, argumentative content rather than generic overviews. The "not a question" constraint for the hook prevents the most overused AI opening pattern.

**17. Social media content series.** "Create a 5-post LinkedIn series about [topic]. Each post should: be 150-200 words, start with a strong first line (no emojis, no 'I' as the first word), include a concrete example or data point, end with a thought-provoking statement (not a question). The series should progress from [basic concept] to [advanced insight] across the 5 posts." — The progression structure turns isolated posts into a coherent content strategy. The specific prohibitions prevent common AI-generated social media clichés.

**18. Content repurposing.** "Here is a [long-form content piece — e.g., blog post, report, presentation]: [paste]. Repurpose this into: 1) A 280-character summary, 2) A 150-word LinkedIn post, 3) A 5-bullet email newsletter snippet, 4) 3 pull quotes for social media, 5) A 50-word meta description for SEO. Maintain the core argument across all formats." — Providing the source material and specifying exact formats for each output is dramatically more effective than asking the AI to "repurpose this content."

**19. Case study from raw information.** "Write a case study based on these details: Client: [description]. Challenge: [what they faced]. Solution: [what we did]. Results: [outcomes/metrics]. Format: 400 words, structured as Situation / Approach / Results / Key Takeaway. Write in third person. Lead with the result in the opening line." — "Lead with the result" is a sophisticated content instruction that produces more compelling case studies by front-loading the proof of value.

**20. Email newsletter edition.** "Write an edition of our newsletter about [theme]. We need: a subject line (under 50 characters, curiosity-driven), a 2-sentence intro that frames why this topic matters now, 3 content blocks (each with a bold heading and 80-100 words), and a sign-off with a call to action for [desired action]. Audience: [description]. Tone: knowledgeable but not academic." — The granular specification of each newsletter component ensures a complete, formatted draft rather than a wall of undifferentiated text. Use the [Prompt Grader](/tools/prompt-grader) to evaluate and refine prompts like these for maximum effectiveness.

Data, coding, and technical prompts (21-30)

**21. Data cleaning instructions.** "I have a CSV with [describe columns]. The data has these problems: [list issues — duplicates, missing values, inconsistent formats]. Write a step-by-step data cleaning plan, then provide the Python/pandas code to execute each step. Add comments explaining what each code block does." — Asking for the plan before the code ensures the AI thinks through the logic rather than jumping to implementation.

**22. SQL query from plain English.** "Write a SQL query for a [database type] database. Tables: [describe schema]. I need: [describe the data you want]. Include comments explaining each JOIN and WHERE clause. Optimise for readability over performance." — Specifying "readability over performance" is important for learning and code review contexts.

**23. Code review.** "Review this code: [paste code]. Evaluate: correctness (does it do what it claims?), readability (would a new team member understand it?), edge cases (what inputs would break it?), performance (any obvious inefficiencies?). List issues by severity: critical, moderate, minor. For each issue, explain why it matters and suggest a fix." — The severity classification makes the review actionable. Without it, all suggestions feel equally important.

**24. Regex pattern.** "Write a regex pattern that matches [describe what you need to match]. Provide 5 test strings that should match and 5 that should not. Then explain the pattern character by character." — The test strings and character-by-character explanation serve as built-in verification and documentation.

**25. API integration plan.** "I need to integrate [API name] into [my application/system]. The use case is [describe]. Outline: 1) Authentication approach, 2) Key endpoints I will need, 3) Data mapping between their schema and mine, 4) Error handling strategy, 5) Rate limiting considerations. Then provide a code skeleton for the core integration in [language]." — This structured approach prevents the common mistake of diving into code before planning the integration architecture.

**26-30: Additional technical prompts.** "Debug this error: [paste error + code]. Walk through the problem step by step before suggesting a fix." | "Write unit tests for [function/module]. Cover: happy path, edge cases, error conditions. Use [testing framework]. Aim for 90% branch coverage." | "Convert this [language A] code to [language B], maintaining the same logic and variable naming. Flag any language-specific patterns that should be adapted rather than directly translated." | "Generate a database schema for [application]. Include: tables, relationships, indexes, constraints. Explain your normalisation decisions." | "Write documentation for this function: [paste code]. Include: description, parameters, return value, exceptions, usage example, and edge case notes." For comprehensive AI-assisted development workflows, explore the [coding curriculum](/coding) which covers these techniques and more.

Strategy and business prompts (31-40)

**31. SWOT analysis.** "Conduct a SWOT analysis for [company/product/initiative] in the context of [market/situation]. For each quadrant, provide 5 specific, evidence-based points (not generic statements). Then identify the single most important strategic implication from the analysis." — "Not generic statements" is the key instruction. Without it, SWOT outputs are filled with platitudes like "strong brand" and "competitive market."

**32. Decision framework.** "I need to decide between [Option A] and [Option B] for [context]. Evaluate each option across these criteria: [list 4-6 criteria]. Score each criterion 1-5 and provide justification. Then give a recommendation with the key risk to monitor for the recommended option." — The structured scoring framework forces rigorous comparison rather than the vague "on one hand, on the other hand" analysis that AI defaults to.

**33. Business case draft.** "Write a business case for [initiative]. Investment required: [amount/resources]. Include: Executive Summary, Problem Statement, Proposed Solution, Cost-Benefit Analysis (with a 3-year horizon), Risk Assessment, Implementation Timeline, and Success Metrics. Audience: [who will approve this]. Assume they are sceptical but fair." — "Sceptical but fair" is a powerful tone instruction that produces realistic business cases rather than overly optimistic ones.

**34. OKR development.** "Help me write OKRs for [team/function] for [time period]. Our strategic priorities are: [list]. For each objective, write 3-4 key results that are: measurable, time-bound, ambitious but achievable, and within the team's control. Then flag any key results that might create perverse incentives." — The "perverse incentives" instruction is sophisticated and catches misaligned metrics that could drive the wrong behaviour.

**35. Stakeholder communication plan.** "I am launching [initiative] that affects these stakeholder groups: [list]. For each group, define: their primary concern, the key message they need to hear, the best communication channel, the timing relative to launch, and who should deliver the message." — This prompt demonstrates a powerful AI use case: systematising complex, multi-audience communication planning.

**36-40: Additional strategy prompts.** "Write a 90-day action plan for [goal]. Break into 3 phases of 30 days each. Each phase needs: objectives, key activities, milestones, and dependencies." | "I'm presenting [recommendation] to [audience]. Anticipate the 5 toughest questions they will ask and draft strong responses for each." | "Analyse this pricing strategy: [describe]. Identify: the value metric, price sensitivity indicators, competitive positioning, and 3 potential risks. Suggest one alternative pricing model to test." | "Draft a post-mortem for [project/initiative]. Structure: what we set out to do, what actually happened, what went well (3), what went poorly (3), root causes, and specific changes for next time." | "Create a change management plan for introducing [AI tool/process] to a [size] team. Cover: stakeholder analysis, communication plan, training approach, resistance mitigation, and success metrics." The principles behind these prompts are taught systematically in the [School of Enigmatica](/school), starting with the [Essentials level](/school/essentials) on prompt engineering fundamentals.

HR, sales, and customer service prompts (41-45)

**41. Job description from scratch.** "Write a job description for [role title] at [company type]. The role reports to [whom] and is responsible for [key responsibilities]. Must-have skills: [list]. Nice-to-haves: [list]. Compensation range: [range]. Write in a direct, jargon-free style. Include a 'What success looks like in 6 months' section instead of a generic responsibilities list. Do not use the phrase 'fast-paced environment' or 'rockstar.'" — The "what success looks like" reframe produces dramatically more useful job descriptions than traditional responsibility lists. The explicit jargon prohibitions prevent the clichés that repel good candidates.

**42. Interview questions by competency.** "Generate interview questions for a [role] position. Organise by competency: [list 4-5 competencies]. For each competency, provide: 2 behavioural questions (STAR format), 1 situational question, and 1 red-flag response to watch for. Difficulty level: [junior/mid/senior]." — Specifying the questioning technique (behavioural, situational) and including red flags transforms this from a generic question list into a structured interview guide.

**43. Sales discovery call preparation.** "I have a discovery call with [prospect role] at [company type/size]. They reached out because of [trigger/context]. Prepare: 10 discovery questions organised from broad to specific, 3 likely objections with response frameworks, a value proposition statement tailored to their likely priorities, and 3 potential next steps ranked by commitment level." — The "broad to specific" question organisation mirrors professional sales methodology, and ranking next steps by commitment level is a sophisticated touch.

**44. Customer service response templates.** "Create response templates for these customer scenarios: [list 5-7 scenarios]. Each template should: acknowledge the customer's situation specifically, explain what we will do (with timeline), and end with a clear next step. Tone: empathetic but efficient. Include [variables] where personalisation is needed. Flag any scenario where escalation to a manager should be offered proactively." — The escalation flagging instruction shows awareness of customer service workflow realities that most AI prompts miss.

**45. Performance review draft.** "Help me write a performance review for [role]. This person's key achievements this period: [list]. Areas for development: [list]. Overall performance: [meets/exceeds/below expectations]. Write the review in a balanced, specific style. Use concrete examples rather than vague praise. Development areas should be framed as growth opportunities with actionable next steps, not criticisms." — "Concrete examples rather than vague praise" and "growth opportunities, not criticisms" produce reviews that are genuinely useful for the employee. These prompt patterns can be explored further in the [Prompt Template Library](/tools/prompt-library).

Creative and advanced prompts (46-50)

**46. Brainstorming with constraints.** "Generate 20 ideas for [challenge/opportunity]. Constraints: [budget, timeline, resources, audience]. First, generate 15 practical ideas that could be implemented within the constraints. Then generate 5 'wild card' ideas that break one constraint each — label which constraint each one breaks and what it would take to remove that constraint." — Separating practical and wild card ideas prevents the common AI brainstorming failure of either all-safe or all-impractical suggestions. The "which constraint it breaks" analysis turns creative ideas into actionable proposals.

**47. Scenario planning.** "We are planning for [situation/decision]. Generate 3 scenarios: 1) Base case (most likely), 2) Optimistic case (things go better than expected), 3) Pessimistic case (key risks materialise). For each scenario: describe the conditions that would cause it, the impact on our [key metric], the leading indicators we should watch for, and the response plan we should prepare. Assign rough probability to each scenario." — This prompt turns the AI into a strategic planning partner. The "leading indicators" element is particularly valuable — it creates an early warning system.

**48. Workshop or presentation design.** "Design a [duration] workshop on [topic] for [audience size and description]. Include: learning objectives (3-4), agenda with time allocations, activities for each segment (mix of presentation, discussion, and hands-on), materials needed, and a participant handout outline. The workshop should be interactive — no segment of pure presentation should exceed 15 minutes." — The 15-minute presentation cap forces interactive design, which is both better pedagogy and better audience engagement.

**49. Process documentation from expert knowledge.** "I'm going to describe a process I do regularly. Ask me questions one at a time to understand every step, decision point, and exception. After you have enough information, write a standard operating procedure with: numbered steps, decision trees for branches, tips and warnings, and a troubleshooting section for common issues. Start with your first question." — This conversational approach to documentation is one of AI's most underrated capabilities. It mimics how a technical writer would interview a subject matter expert.

**50. Meta-prompt: prompt improvement.** "Here is a prompt I have been using: [paste prompt]. Analyse it for: clarity (will the AI understand exactly what I want?), completeness (is any critical context missing?), specificity (are the output format and quality expectations clear?), and efficiency (could it achieve the same result with fewer tokens?). Then rewrite the prompt with improvements and explain what you changed and why." — This meta-prompt is how you improve at prompt engineering over time. Run your most-used prompts through this analysis quarterly. For a more structured approach, the [Prompt Grader tool](/tools/prompt-grader) provides automated prompt quality assessment, and the [CONTEXT Framework](/context-framework) gives you a systematic methodology for crafting effective prompts from scratch.

Related Terms