free-ai-lab free-ai-lab

AI Prompts 101: Learn Prompting from Zero

What is an AI prompt, how do you use one, and where do you paste it after copying? A zero-to-one beginner guide: the three elements, one universal formula, and the most common mistakes.

Read in other languages: 中文

In one sentence: a prompt is simply the text you type to an AI — a good prompt gets great work; a vague one gets filler.

This guide is for complete beginners. By the end, you’ll be able to pick any prompt from the Prompt Library and use it right away.

Step 1: What exactly is a prompt?

Analogy: an AI is like a smart intern with no work experience. A prompt = the task brief you give that intern. The clearer the brief, the better the work.

ApproachYour promptHow the AI performs
❌ Vague“Write something for me”No idea what you want — generic output
✅ Clear“Write a 300-word social media post introducing a portable juicer cup, with 3 emojis and 5 hashtags at the end”Right on target, barely needs edits

Step 2: The three elements of a good prompt

  • Role: “You are a social media copywriter with 3 years of experience” — set the AI’s identity and it writes in that voice
  • Task: “Write a recommendation post” — say exactly what you want done
  • Requirements: “300 words, with emojis, 5 hashtags at the end” — length, format, and style are all requirements

Every prompt in our library is pre-written with “role + task + requirements” — just copy, swap the variables, and go.

Step 3: Where do you paste it? (three uses)

Chat tools (easiest): ChatGPT, Claude and similar chat tools: hit copy → paste into the chat box → enter. The AI does the rest.

Image tools: conversational image generators like ChatGPT: paste the prompt → get an image. Ready-made image prompts with sample images live in the Image Gallery.

Coding tools: Cursor, VS Code and similar tools can inject prompts as “system prompts” or project rule files — more on that in the advanced guide.

Step 4: The 4 mistakes beginners make most

  • Too vague: “make it look better” — turn it into measurable requirements (font, colors, style words)
  • Cramming everything in at once — one goal per request; split multi-goal work into separate asks
  • Giving up after one try — it’s normal for the first output to miss; just reply with clarifications (chat tools support continuous follow-ups)
  • Assuming you must write prompts in English — modern AI tools handle your native language fine; don’t let English scare you off

Next: try one