Prompting Fundamentals

Every topic in this track has been quietly building toward one sentence. Here it is:

The prompt is not a command. It's the opening of a document, and the model's only job is to finish it plausibly.

That's the whole of prompting. Everything below is a consequence.

There is no listener in there. Nothing reads your request, forms an intention, and sets out to satisfy it. There is a text-continuation engine — the same one from the very first topic — and you are handing it the first page.

Which means the real question is never "how do I ask for what I want?" It's:

What kind of document has my answer as its natural continuation — and did I write the opening of that document?

The two openings

Open like this:

write about our product launch

What plausibly follows? Text that continues a vague, contextless, low-effort line. So: something generic, hedged, four bullet points and a summary. The model did not fail you. It finished the document you actually started.

Now open like this:

You're writing the internal launch note for Retain, a learning app. Audience: 40 engineers who already know the product. Two paragraphs, no marketing voice, lead with what changed for them. Release notes below.

Now the plausible continuation is the thing you wanted — because you wrote the opening of a document whose natural continuation could hardly be anything else.

You didn't ask harder. You made your answer the likely one. That is the entire skill.