What is meta prompting?

Meta prompting is using an AI model to write or improve prompts instead of writing them all yourself. You describe the job you want done, and the model drafts, critiques, or rewrites the prompt for you. It also refers to a related idea: writing prompts that focus on structure and format — the shape of a good answer — rather than piling in specific examples.

How does meta prompting actually work?

In practice it looks like a short conversation. You tell the model what you're trying to get, and ask it to produce the prompt. A few common moves:

• Generate — "Write me a prompt that turns meeting notes into a clean action-item list."
• Improve — paste a prompt that isn't working and ask, "Why might this give weak results, and rewrite it."
• Template first — ask for the ideal structure of an answer (headings, order, tone) before you ever fill in the details.

The second sense of the term matters too: a meta prompt cares more about the format of a good response than the exact content. You define the skeleton, and the model handles the specifics each time you reuse it.

Why is it useful?

Most people are better at recognizing a good prompt than writing one from scratch. Meta prompting shifts the hard part onto the model, which is genuinely good at spelling out instructions, edge cases, and structure you'd forget. It's a fast way to get a strong first draft, and it works well when you're building a prompt you plan to reuse many times.

How is it different from prompt chaining?

They sound similar but solve different problems. Meta prompting is about creating one better prompt. Prompt chaining is about running several prompts in sequence, where each step's output feeds the next.

You can use both together — meta prompt each individual step, then chain those steps into a workflow.