Skip to lesson content
Free course

AI Foundations / Module 4 / M04-U00 · 4-5 minutes

M04-U00 · 4-5 minutes

Welcome to Module 4: Define the Work Clearly

Work through the explanation, apply it to the example, and complete the quick check before continuing.

Compare two requests

Request A says:

Write something about our workshop.

Request B says:

Write a 120-word email for adults who registered for Saturday's beginner photography workshop. Use only the approved event details below. Confirm the date, time, place, and items to bring. Use a warm, direct tone. Do not add a price, instructor biography, or parking promise. End with the supplied contact address. Before finishing, check every factual detail against the source notes.

Request B is longer because it defines the work. It tells us the desired result, audience, source, limits, format, tone, and checks.

This fuller description is a task specification. A prompt may contain the specification, or it may point to files and instructions that form the specification together.

Clear instructions do not guarantee a perfect result

A strong specification improves the chance of a useful output. It does not make a model perfectly reliable.

The source may be incomplete. The model may miss a constraint. The task may need a different tool. A person still has to review according to the risk.

The purpose of a specification is not to control every word. It is to make the result easier to produce, inspect, correct, and approve.

What you will learn

You will learn to define:

  • Outcome and audience.
  • Relevant context and authorised source material.
  • Constraints, exclusions, and boundaries.
  • Output format and review criteria.
  • Useful examples and clarification questions.
  • Stages for complex work.
  • Targeted changes after diagnosing a weak output.

Our main rule

Improve the description of the work, not the decoration around the prompt.

There is no magic phrase that removes the need for clear requirements and review.

At the end of the module, you will turn three vague requests into usable specifications. Let us begin with the most important question: what should the result achieve, and for whom?