Prompts IA sélectionnés pour des usages professionnels, créatifs et techniques.
Works well with Claude Opus 4.8, GPT-5.4, and Gemini 3 Pro — needs strong instruction-following and a large context window to hold pasted email threads. Claude and GPT-5.4 handle the scoring rubric most consistently; Gemini 3 Pro is a solid free-tier alternative for smaller batches.productivityYou're back from a 3-day conference with 140 unread emails and 45 minutes before your next meeting. Somewhere in that pile is a client threatening to cancel, a deadline moved up, and a dozen threads that can wait until next week — but you can't tell which is which without reading every single one.
Un Prompt structuré qui note chaque e-mail par urgence et impact, puis vous remet une liste d'actions classée avec des réponses rédigées pour tout ce qui est urgent.
prompt-engineeringemail-triage
Works best with GPT-5.4 or Claude Opus 4.8 for the trigger labeling accuracy. Gemini 3.5 Flash and DeepSeek V4 also work well for the initial hook generation.marketingYou're launching a cold email campaign or paid ad set for a product next Monday, and you have one afternoon to generate enough genuinely different headline angles to A/B test properly — instead of writing three headlines from gut instinct and hoping one works.
Un prompt qui génère 20 variantes de titres vraiment différentes pour n'importe quelle campagne, chacune étiquetée avec le déclencheur psychologique spécifique qu'elle utilise — conçu pour de vrais tests A/B, pas pour des conjectures.
marketingcopywriting
Works best with GPT-5.4 or Claude Opus 4.8. Gemini 3.5 Flash and DeepSeek V4 also handle this well, though they occasionally need a reminder in the prompt to avoid hypothetical phrasing.Startups et EntreprisesYou believe freelance graphic designers are losing money to scope creep, and you've booked eight customer interviews for next week before writing a line of code — you need questions that will actually surface whether the problem is real and costly, not questions that make you feel good about your idea.
Un prompt qui génère des questions d'entrevue conçues pour attraper la validation de faux positif qui tue les startups avant leur lancement — pas de questions orientées, pas d'hypothèses, seulement des comportements passés.
startupsproduct-management
Works best with Claude Opus 4.8 or GPT-5.4 for large response sets (500+ rows); Claude Sonnet 5 and Gemini 2.5 Pro handle smaller surveys (under 200 responses) equally well.Analyse de DonnéesA People Ops manager just closed a 340-response employee engagement survey across the engineering organization and has 48 hours to present findings to the VP of Engineering before a headcount planning meeting, with only a raw CSV export and no built-in analytics dashboard.
Un prompt structuré qui transforme les exports bruts d’enquête en un rapport de résultats — ventilation quantitative, thèmes codés avec citations, analyse de sentiment, et actions recommandées spécifiques, pas seulement un résumé de moyennes.
prompt-engineeringsurvey analysis
Works best with Claude Opus 4.8 or GPT-5.4 (large context window handles long reports cleanly). Claude Sonnet 5 and Gemini 2.5 Pro also perform well for reports under 15 pages.Rédaction et CommunicationA chief of staff receives a 34-page market research report from an outside consulting firm at 4 PM, with a board meeting scheduled for 9 AM the next morning where the CEO needs to speak to its findings with confidence.
Un prompt structuré qui transforme n’importe quel long rapport en un brief exécutif d’une page — conçu pour faire ressortir exactement ce qui change une décision, et non pour tout résumer.
prompt-engineeringproductivity
Claude Opus 4.7 (also works well with GPT-5.4 and Gemini 3 Pro)CarrièreYou found a job posting that's a strong match for your background, but you're not sure if your current resume — written for a previous, slightly different role — actually surfaces the right keywords and accomplishments, and you have limited time before the application deadline to figure out what to fix.
Un prompt qui compare votre CV à une offre d’emploi précise, signale les lacunes qui vous feraient écarter par un ATS ou un recruteur, et réécrit vos points faibles en utilisant la formule XYZ pour qu’ils correspondent exactement à ce que le poste exige.
Careerresume
Claude Opus 4.7 (also works well with GPT-5.4 and Gemini 3 Pro)Finance et NégociationYou just received a job offer with a base salary about 8% below the market rate you researched on Levels.fyi and in your network, and you have a call scheduled with the recruiter in two days to discuss compensation before you can accept.
Un prompt structuré qui prend votre rémunération actuelle, la nouvelle offre et les données de marché pour produire un script de négociation complet — chiffre d’ancrage, phrases d’ouverture, réponses aux objections et votre point de rupture — pour aborder l’appel préparé plutôt que d’improviser.
Careersalary-negotiation
Claude Opus 4.8 (also works well with GPT-5.4 and Gemini 3 Pro for most languages; for niche frameworks, verify generated syntax against current library docs)Outils de DéveloppementYou just wrote a pricing function for checkout and need a real test suite before it ships, but you have 15 minutes before a deploy window and writing every edge case by hand from scratch would take an hour.
Un prompt qui génère une suite de tests complète — chemin nominal, cas limites, scénarios d'erreur et lacunes d'intégration — pour n'importe quelle fonction, plus une liste honnête de ce qu'il n'a pas pu déterminer sans plus de contexte.
testingprompt-engineering
Claude Opus 4.8 (also works well with GPT-5.4 and Gemini 3 Pro — any model with strong instruction-following and a large context window for long documents)Outils de DéveloppementYour engineering team just shipped a new rate-limiting change to the API, and your account executive has a call with a Fortune 500 procurement team in two hours who will ask what this means for their integration timeline and reliability.
Un prompt qui réécrit la documentation technique pour des parties prenantes non techniques sans sacrifier la précision — plus un glossaire et des réponses préconstruites aux questions qu'ils poseront ensuite.
prompt-engineeringtechnical-writing
Claude Opus 4.8 or GPT-5.6 with web search/browsing enabled (verification quality depends heavily on the model being able to check claims against live sources — without search access, treat the output as a risk triage, not a ground-truth check)Intelligence ArtificielleYou used an AI assistant to draft a client-facing quarterly report that includes market statistics, a competitor comparison, and two quotes attributed to industry analysts, and you have 15 minutes before the file needs to go out to review it for anything that could be wrong or made up.
Un passage de relecture structuré qui signale les statistiques fabriquées, les citations inventées et les références invérifiables dans un texte généré par IA avant qu'il n'atteigne un client, une publication ou un dépôt juridique.
fact-checkinghallucination
Claude Opus 4.8 (also works well with GPT-5.6 and Gemini 3.5 Pro for structured data reasoning; avoid smaller/faster model tiers for this one — it needs multi-step causal reasoning, not just summarization)Analyse de DonnéesYou're a growth PM at a 40-person B2B SaaS company. In Monday's leadership meeting, your CEO asks why the April cohort of new signups is retaining 12 percentage points worse than March's cohort at the Week 4 mark, and you have 20 minutes before your next meeting to come back with an answer that isn't 'we're not sure yet.'
Transformez un export brut de données de rétention en hypothèses classées et testables sur les raisons pour lesquelles une cohorte a un churn plus élevé que les autres — en minutes au lieu de semaines.
product-analyticsretention-analysis
Claude Opus 4.8 (also works well with GPT-5.4; models with strong reasoning depth produce sharper counter-arguments than fast/small models, which tend to default to generic "some critics say" hedging)learningYou're a startup founder about to pitch investors on doubling down on a paid acquisition channel that has driven 80% of your growth over the past year. Before the meeting, you run this prompt against your own thesis — that concentrating spend in this channel is the right call — to see the strongest case against channel concentration risk, so you walk into the pitch having already answered the hardest question instead of improvising when a partner asks it live.
Un prompt qui construit l'argument opposé le plus fort à tout ce que vous croyez — pas un homme de paille faible — plus les preuves exactes qui prouveraient que vous avez tort, pour ne jamais être pris au dépourvu par la question que vous n'aviez pas préparée.