Skip to course content
Free prompt workflow course

Prompting and AI Workflow Design

Unit ID: PAW-M02-U02 Estimated active time: 15-20 minutes

What should the output help you do?

Before asking for output, ask:

> What decision or action should this output support?

This question makes the task sharper.

AI output should not just exist. It should help someone do something.

Examples of output decisions

An output may help you:

Each decision needs a different output.

Weak versus clear

Weak:

> Compare these ideas.

Clearer:

> Compare these three fictional workshop ideas so I can choose one for a beginner audience. Use criteria: learner value, effort, risk, and whether it can be done without paid tools.

The second version tells the AI tool what the comparison is for.

Decision does not always mean final answer

Sometimes the output should not make the final decision.

It may only support the decision.

For example, you can ask:

> List questions I should answer before choosing.

Or:

> Compare options, but do not recommend one yet. First show assumptions and missing information.

This is useful when the task is uncertain.

Ask for decision support, not false certainty

AI tools often sound confident.

If the task has missing facts, ask the tool to show:

This keeps the output honest.

Example

Job statement:

> I need to prepare a meeting agenda so that the team can agree on next steps.

Output decision:

> The agenda should help the team decide who owns each next action.

Now the prompt can ask for:

Mini practice

For each job statement, write the output decision.

  1. I need to create a study checklist so that a beginner learner can prepare for a quiz.
  2. I need to draft a polite update so that a fictional customer understands the next step.
  3. I need to compare three safe project ideas so that a small team can choose one.

Use this pattern:

> The output should help [person/group] [decide/do/check/prepare].