Curated AI prompts for business, creative, and technical use cases.
Works well with Claude Opus 4.8 or GPT-5.4 for nuanced tone-matching across channels; Claude Sonnet 5 handles shorter transcripts fine at lower cost.marketingYou just finished five 30-minute customer interviews for a case study project. You have five raw transcripts, roughly 45 minutes of usable content buried in filler words, tangents, and repeated phrasing, and marketing needs three strong quotes for a landing page redesign by Friday.
Un prompt qui prend des transcriptions d’entretiens clients désordonnées, des notes d’appels de support ou des textes d’avis, et extrait des citations de témoignages prêtes à l’emploi, organisées par l’objection ou le cas d’usage spécifique que chacune aborde.
prompt-engineeringmarketing
Works best with Claude Opus 4.8 or GPT-5.4 for the reasoning depth needed to rank causes by plausibility; Claude Sonnet 5 is a solid faster/cheaper alternative for routine metric checks.Data AnalysisYour weekly signup dashboard shows conversions dropped 18% starting Tuesday. You don't know if it's a pricing page bug, a marketing campaign that ended, a competitor promotion, seasonality, or a tracking error — and your team is about to spend the next two days randomly checking each one.
Un prompt structuré qui prend une anomalie métrique (pic, chute ou plateau dans n'importe quel KPI) plus le contexte de vos systèmes et des changements récents, et renvoie une liste classée et pondérée par des preuves des causes probables avec une vérification spécifique suivante pour chacune.
prompt-engineeringdata analysis
Claude Sonnet 5 (also works well with GPT-5.4 and Gemini 3 Pro — any model that follows structural constraints reliably)Developer ToolsAn on-call engineer has a payments API outage from 2am with only scattered Slack timestamps and half-remembered details, and needs to turn that into a clear, professional postmortem before Monday's leadership review — without spending two hours writing it from scratch or accidentally writing something that reads like it's blaming a specific teammate.
Un prompt structuré qui transforme des notes d'incident désordonnées en un post-mortem professionnel et sans reproche — avec une chronologie réelle, une cause racine clairement identifiée, et pas plus de cinq actions correctives assignables.
incident-responsesre
Claude Sonnet 5 (also works well with GPT-5.4 and Gemini 3 Pro — any model with strong structured reasoning)learningA backend engineer with four years of Python and PostgreSQL experience wants to move into a Machine Learning Engineer role within four months, has 8 hours a week to study, and has no idea whether to start with math fundamentals, a specific framework, or a portfolio project first.
Un prompt structuré qui cartographie l’écart exact entre vos compétences actuelles et un rôle cible, puis élabore un plan d’apprentissage semaine par semaine avec des jalons concrets – pas juste une liste de lectures.
career-developmentlearning-plan
Claude Opus 5 (also works well with GPT-5.4 and Gemini 3 Pro)creativeYou have 4 minutes on the agenda at next week's board meeting to explain why the company's new fraud-detection model reduces false declines, and the board has finance and legal backgrounds with zero machine learning experience — a slide full of precision and recall numbers will lose them in the first 20 seconds.
Un prompt qui génère cinq métaphores visuelles distinctes et testées auprès du public pour tout concept complexe, avec des visuels de diapositives, des répliques parlées et une évaluation des risques pour chacune — pour que vous n'ayez pas à deviner quelle analogie fonctionnera avec votre public spécifique.
communicationpitch-deck
Claude Opus 5 (also works well with GPT-5.4 and Gemini 3 Pro — any model with a large context window for pasting full contract text)finance-negotiationYour company's project management SaaS vendor sends a 22-page renewal agreement three business days before the auto-renewal deadline. Legal review normally takes a week, but you need to know today whether anything in the contract needs to be pushed back on before you sign or let the renewal window close.
Un prompt qui lit un contrat fournisseur, signale les clauses qui créent réellement un risque, et rédige le langage de redline exact ainsi que les points de négociation pour chacune — au lieu d'un résumé générique « lisez ce contrat ».
Negotiationlegal risk
Works with any capable model with a large context window (Claude Sonnet/Opus, GPT-5 tier, Gemini 3 Pro) — needs enough context length to hold a full meeting transcript, so avoid small/fast tiers for long recordingsWriting & CommunicationA product manager just left a 45-minute cross-functional planning meeting between engineering, design, and marketing about a delayed Q3 feature launch, and needs to send a decision recap within the hour — before people's memories of who-agreed-to-what start to diverge — but manually re-listening to the recording or notes to reconstruct exactly what was agreed would take 20+ minutes and still risk missing a contradiction between two speakers.
Arrêtez de laisser le « on verra plus tard » être consigné dans un compte rendu comme une décision ferme. Cette consigne sépare ce qui a réellement été convenu de ce qui a simplement semblé l’être.
prompt-engineeringproductivity
Works with any frontier reasoning model (Claude Opus/Sonnet, GPT-5 tier, Gemini 3 Pro) — pair with a quick web check for current per-token pricing, since the model's training data may lag real-time rate changesArtificial IntelligenceA support-team lead needs to route 40,000 monthly customer tickets into 12 categories with a suggested first-response draft, must stay under 3 seconds per ticket, keep costs under $2,000/month, and cannot send EU customer PII outside the EU — but has no AI infrastructure background to know which model tier fits those constraints without expensive trial and error.
Arrêtez de recourir systématiquement au modèle le plus cher pour chaque tâche. Ce prompt construit une comparaison coût/latence/qualité entre les différents niveaux de modèles et vous fournit une recommandation défendable, pas une simple supposition.
prompt-engineeringinfrastructure
Claude Opus 4.7 (works well with GPT-5.4 and Gemini 2.5 Pro; needs a model with strong structured-table output)Startups & BusinessA 12-person SaaS company just wrapped a quarterly strategy offsite with a completed SWOT analysis on a whiteboard — churn is flagged as a weakness, rising AWS costs as a threat, a competitor's weak enterprise tier as an opportunity — but the founder has to walk into Monday's all-hands with a concrete 30-day plan, not a photo of a whiteboard nobody will look at again.
Un prompt qui prend une analyse SWOT complétée et la convertit en un plan d'action de 30 jours semaine par semaine avec des responsables, des échéances et des indicateurs de succès — pour que la session de stratégie change réellement ce que votre équipe fait le lundi.
prompt-engineeringstartup-strategy
GPT-5.4, Claude Sonnet 5, or any reasoning-capable model — works best with one that can hold both weeks' context and reason about relative priority, not just summarizeproductivityIt's Monday morning, you have 15 minutes before your first meeting, and you have a messy calendar export plus a half-remembered sense of what mattered last week — you need an honest read on where your time actually went, not just a to-do list for the next five days.
Transformez une semaine chaotique de tâches et de réunions en un bilan honnête par rapport à vos vraies priorités, et une liste classée pour la semaine à venir – pas juste une énième liste de tâches.
prompt-engineeringproductivity
GPT-5.4 or Claude Sonnet 5 — any frontier model with a large context window handles this well; feed it the full article text, not a summarymarketingYou just published a 1,500-word blog post analyzing a major industry report, and your social calendar for the week is due in 20 minutes — you need 10 genuinely different posts, not the same headline copy-pasted across five platforms.
Transformez n'importe quel article de blog en 10 publications sociales véritablement distinctes et natives de chaque plateforme en une seule passe — pas le même titre copié-collé cinq fois.
prompt-engineeringsocial-media
Works best with Claude Opus 4.8 or Sonnet 5 for large functions (context window and code reasoning matter more than raw speed here); GPT-5.4 is a solid alternative. Avoid smaller/faster models for this — risk scoring requires holding the whole function's logic in view at once.Developer ToolsYou've inherited a 400-line function that everyone on the team avoids touching because it 'works, mostly' and nobody remembers why it's structured the way it is. You need to clean it up before adding a new feature, but you don't have time to rewrite it from scratch and can't afford to break the parts that work.
Un prompt structuré qui examine une fonction connue comme problématique, propose des refactorisations spécifiques avec code avant/après, et attribue à chacune un score de risque de régression avant que vous ne touchiez à quoi que ce soit.