What is a prompt?
A prompt is the text you give an AI model, such as ChatGPT or Gemini, to tell it what to do. It can be a question, an instruction, an example or context about your situation. The clearer and more specific the prompt, the closer the answer gets to what you wanted.
Updated:
Types of prompts
Google's machine learning glossary defines a prompt as any text entered as input to a large language model to condition it to behave in a certain way. A prompt can be as short as a phrase or as long as a whole novel. There are several types:
- Question: "What is LinkedIn SSI?"
- Instruction: "Write a 150-word post about..."
- Example: an instruction followed by an example of the result you want
- Role: "Explain this to a CEO with no technical background"
Writing prompts is called prompt engineering, or prompt design. Google's Gemini guide recommends clear and specific instructions, examples, context, and breaking a big task into parts.
Example: a weak prompt vs a strong one
Weak: "Write a LinkedIn post about our service."
Strong: "Write a 150-word LinkedIn post for finance leaders at software companies. Open with a problem they know from closing a quarter. Use this example from my week: [...]. No superlatives, no hashtags."
The second prompt gives an audience, a goal, a length, real material and constraints. The result will be far less generic.
Common confusion
Many people think a perfect prompt will make the text sound exactly like them. In practice, a general model doesn't know your voice, your stories or your opinions. A good prompt improves the result, but it doesn't replace real material.
How it relates to Velut
That is exactly why Velut AI, our separate product, trains a personal model on your own LinkedIn writing instead of relying on a prompt alone. For managed clients, I write by hand. There is also another side: prompts are how buyers ask ChatGPT about vendors in your category. To appear in the answer, you need content the model can cite.
Frequently asked questions
What is prompt engineering?
Prompt engineering is writing and refining prompts to get the answer you want from an AI model. Google also uses the term prompt design as a synonym.
How long should a prompt be?
There is no single right length. A short prompt is enough for a simple question. For a complex task, add context, examples and constraints, even if that takes a few paragraphs.
Can I write prompts in other languages?
Yes. The major models understand many languages, including Hebrew. If the result is off, add an example of the style you want in that language.