Glossary
Prompt
A written instruction given to an artificial intelligence to tell it what is expected. It can be a command, a question, or a guideline expressed in natural language.
You can think of a prompt as a recipe or a creative brief:
- It specifies what you want to get (e.g., a product description, a summary, an image…),
- You can indicate a style, tone, length, or target audience,
- It often includes contextual or reference information.
🎯 Prompt goal: to guide the AI so that it produces a coherent, useful, and tailored response.
The clearer, more structured, and more contextualized a prompt is, the more relevant the generated response will be.
🧭 What is a prompt used for?
A prompt can be used to:
- Generate text (e.g., a summary, product sheet, LinkedIn post, email…),
- Extract information (e.g., identify key points from a document),
- Rewrite or improve content (e.g., correct style or adapt to an audience),
- Create code, images, ideas…
💡 Best practices for writing an effective prompt
A well-crafted prompt should generally answer the following questions:
- Who is speaking? → the role the AI should play.
- What is the task? → the main expected action.
- What is the subject? → the data or information to process.
- For whom? → the target audience.
- How? → tone, style, format, constraints.
✅ Example of a simple and well-structured prompt
🎯 Goal: generate a marketing description for an e-commerce product page
🖋️ Prompt:
You are a marketing copywriting expert specialized in e-commerce. Your mission is to write a clear, engaging, and conversion-oriented product description. Product to describe: “SmartTime Pro 2” connected watch. Main features: 1.5” AMOLED display, sleep tracking, Bluetooth calls, 10-day battery life, compatible with iOS/Android. Target audience: active adults aged 25–45, living in urban areas, interested in technology and well-being. Style constraints: - Tone: dynamic, reassuring, and persuasive. - Style: fluid and jargon-free. - Format: continuous text of 3 to 5 sentences max (no bullet points). - Goal: quickly convince a potential customer on a product detail page.
LLM
An LLM, or Large Language Model, is an artificial intelligence specialized in processing human language.
Token
Un token est une unité de texte utilisée par les modèles de langage pour "lire" et "comprendre" les mots. Un token n’est pas forcément un mot entier, c’est plutôt une portion de texte (un mot, une syllabe, ou un caractère).