# Writing a prompt for an AI: getting a usable result the first time URL : https://skillrung.com/en/courses/artificial-intelligence/writing-a-prompt-that-works Topic: Artificial intelligence · Duration: 25-30 min · Updated on 2026-09-15 ## In brief A prompt that works has five parts: the role you are addressing, the useful context, the task, the constraints and the expected format. Leaving one out means letting the assistant decide on its own. Everything that matters has to fit inside the current conversation: as soon as the topic changes, open a new one. If the result disappoints, do not start again from scratch: name the part that was missing and ask again. ## Programme - 1. Why the result comes out wide of the mark : Understand where a disappointing result comes from and what is at stake in the wording - 2. Write a prompt that works : Build a complete prompt and follow up usefully when the result is not right - 3. Your action plan : Turn the five parts into a habit in the writing you do this week 1 content sections, each with its quiz. Three level quizzes: Beginner, Intermediate, Expert. ## Free preview (first 4 slides) ### 1. It is not the tool that misunderstood You type a ten-word request between two meetings. What comes back is polite, correct and unusable: nothing precise, everything to be rewritten. The reflex is to blame the tool. But you had not said who you were writing to, or why, or in what form. The assistant filled in the gaps for you, and it knows nothing about your structure. ### 2. What you will be able to do by the end - Spot what a prompt is missing before you send it - Write a request the assistant cannot read the wrong way - Follow up on the right point when the answer disappoints ### 3. Four common complaints, four causes in the prompt You have surely said one of these this week. Every time, the cause lies in what was not written. - “It is too long”: no limit had been given - “It looks nothing like us”: no context - “It makes things up”: the prompt did not allow for doubt - “It just agrees with me”: you had asked for a confirmation ### 4. Where do you stand? One question before you read anything at all. It measures the gap between what you think you are asking for and what the assistant actually receives. _Self-assessment of 1 questions in the player._ ## Access The first four slides of every course and the Beginner-level quiz are free, with no account. From the fifth slide onwards a subscription is required: €29.99/month incl. VAT, no commitment, cancellable online. skillrung is not a French training body certified under the Qualiopi scheme; its content is not eligible for the French CPF, OPCO or DPC schemes. ## Frequently asked questions ### Do you have to write a long prompt to get a good result? No. What counts is precision, not length. Three lines that name the reader, the deliverable and the question expected are worth more than a page of generalities. Write what the assistant cannot guess, and nothing else. You will soon know what to add: the first result shows you what you had left to its choice. ### The result does not suit me: do I have to write the prompt again from the start? Rarely. Start from what was produced and say precisely what is wrong: too long, wrong recipient, unsuitable tone, invented passage. A targeted follow-up corrects things faster than a rewritten prompt. On the other hand, if you change topic or if the exchange has dragged on, it is better to open a fresh conversation and put back into it only what is useful. ### Can you reuse the same prompt from one time to the next? Yes, and that is the whole point. A prompt that has worked well is kept in a file, with the passages to be replaced marked. You then only rewrite the context of the day. Reread it before each use all the same: a constraint that was useful last week may no longer match today's need. ## Sources - Règlement (UE) 2024/1689 établissant des règles harmonisées concernant l'intelligence artificielle : EUR-Lex : Journal officiel de l'Union européenne, 2024-07-12, https://eur-lex.europa.eu/eli/reg/2024/1689/oj?locale=fr - Recommandations de sécurité pour un système d'IA générative : ANSSI, 2024-04-29, https://messervices.cyber.gouv.fr/guides/recommandations-de-securite-pour-un-systeme-dia-generative - Les fiches pratiques IA : CNIL, 2026-09-11, https://www.cnil.fr/fr/les-fiches-pratiques-ia - Mata v. Avianca, Inc., opinion and order on sanctions, No. 22-cv-1461 (S.D.N.Y.) : Justia : US District Court, Southern District of New York, 2023-06-22, https://law.justia.com/cases/federal/district-courts/new-york/nysdce/1:2022cv01461/575368/54/