Foundation Path·Input Mechanics·10 min

Prompt Pump

Technique beats force — in prompting too.

Goal

You refine your prompts so Copilot delivers exactly what you need — instead of something vaguely similar.

Why this matters

You know the moment: you make an effort, phrase your prompt carefully — and Copilot delivers something completely different from what you need. Many give up after the second attempt, thinking: “It just doesn’t work.”

But a good prompt is like a clean lift: technique beats force. This session trains the technique.

The drill

  1. Take a prompt you already used this week — for a summary, a proposal or a mail.
  2. Open Copilot and start with this variant:
I want to improve the result. Here is my original prompt:
[your prompt]
What can I do so you understand me better?
What are you missing for a good result?
  1. Build the answer into your prompt and run it again. Compare the two results.

That’s the entire trick: you ask Copilot what it’s missing — instead of guessing.

Extra round

From now on, build your prompts from four blocks:

  • Goal: What should Copilot do?
  • Context: Why do you need the result?
  • Expectation: In what format or tone?
  • Source: Which data or documents should Copilot use?

Take an old prompt and retrofit it with all four blocks. The difference is immediately visible.

Micro-win

Today you didn’t simply ask — you deliberately phrased.

Remember

Every good prompt starts with clarity and structure, not with luck. And clarity about the task is thinking work no tool takes off your hands — Copilot only amplifies it.

Next in the plan

Back to the plan: Foundation Path →

FAQ

How do I improve my Copilot prompts?

Ask Copilot itself: show it your original prompt and ask what it's missing for a good result. The answer is your instruction manual — more precise than any guessing.

What are the building blocks of a good prompt?

Four: goal (what to do), context (why), expectation (format and tone), source (which data). Prompts with all four blocks deliver noticeably more precise results.

Why does Copilot deliver something different from what I wanted?

Almost always because the prompt leaves goal or context open. Copilot fills gaps with assumptions — the clearer the instruction, the smaller the gap.

Is it worth reworking prompts instead of writing new ones?

Yes. A refined prompt shows you which block was missing — that's the learning effect a fresh attempt from zero doesn't deliver.

How do I get permanently better at prompting?

Through the feedback loop: phrase, ask Copilot for gaps, retrofit, compare. Make that loop a habit and eventually you'll barely need it.