Lesson 07 Real-world examples
Theme:

Homework — Real-world examples

Estimated time: ~50 min · All you need: access to Claude, ChatGPT, or any AI chat · No code, no installs

This homework is different from most: every task uses something real from your actual life or work. Not hypothetical scenarios, not test prompts — real tasks. That's the only way to build the instinct. Pick the tasks that are most relevant to you. If you only have time for two, do Tasks 1 and 3.

Before you start

Open Claude (claude.ai), ChatGPT (chat.openai.com), or any AI chat you have access to. Keep the Theory page open in another tab — specifically the Use-Case Explorer, which has ready-to-adapt prompts you can use as starting points.

Task 1 — The force multiplier: one real task from your week

Pick a real task you've been putting off or struggling with this week. It can be from any of the five categories — writing an email, summarising a document, getting unstuck on something creative, understanding a topic, or doing something with code. The only rule: it has to be real, not a practice exercise.

Step A: Do it the old way first (or note how you'd normally do it).

Write down roughly how long it would take you to do this task without AI. What are the hard parts?

Step B: Use AI to do it.

Start with a prompt from the Theory Explorer, or write your own. If the first output isn't quite right, don't give up — iterate. Say "that's close but [specific thing that's wrong] — try again." Two or three exchanges is normal.

Step C: Evaluate what you got.

How does the output compare to what you'd have produced yourself? What did it get right? What did it miss or get wrong? What would you change for the next time you try this type of task?

Self-check

  • Did you use a real task, not a hypothetical?
  • Did you iterate at least once (refine based on the first output)?
  • Can you identify one specific thing the AI did better than you expected, and one thing it did worse?

Task 2 — The before/after: write a messy input, get a polished output

Pick one of these inputs — whichever feels most natural to you:

Then write a prompt that describes exactly what you want to come out of the other end. Include:

Self-check

  • Is the before genuinely raw / imperfect (not a cleaned-up version)?
  • Does your prompt specify audience, tone, and format?
  • Is the after noticeably better in at least two specific ways you can name?
  • What would you change about the prompt if you did this again?

Task 3 — The learning session: AI as tutor for 15 minutes

Pick something you've been meaning to learn or understand better. It can be a professional topic (a technology, an area of business, a skill), a personal interest (history, science, a language, an art form), or something that's come up in your work lately and you want to understand more deeply.

Step A: Start a tutoring session.

Use this prompt as a template:

You are a patient tutor. I want to learn about [your topic]. I'm a [complete beginner / have some basics / intermediate learner]. Start with [the first concept you want to cover]. Explain it simply with a real-world example. After your explanation, ask me one question to check I understood, and wait for my answer.

Step B: Answer the AI's question. Don't look anything up — give your honest first answer. If you're wrong, note the correction. Ask follow-up questions if something is unclear.

Step C: Ask for a quiz. Once you feel you've understood the first concept, say: "Now quiz me: give me 3 questions on what we just covered and grade my answers."

Step D: Reflect. After 15 minutes, write down: what did you learn that you didn't know before? What surprised you? How does this compare to trying to learn the same thing from a book, article, or video?

Self-check

  • Did you choose a real topic you actually want to know more about?
  • Did you get at least two rounds of back-and-forth before the quiz?
  • Did you answer the quiz questions from memory (not googling)?
  • Can you explain the main concept you learned to someone else in plain language?

Task 4 — The creative sprint: brainstorm something you genuinely need

Pick a creative brainstorming task that's actually useful to you right now. Some examples:

Step A: Write a constrained brainstorm prompt. Include:

Step B: React and iterate. After the first batch, don't just pick one. Say: "I like options [X] and [Y] — give me 10 more in that direction." Do at least one round of iteration.

Step C: Pick your actual favourites from the full set and note why you chose them. What quality made them stand out?

Self-check

  • Is this a real brainstorming need, not a hypothetical exercise?
  • Did your prompt include exclusions or constraints that narrowed the output usefully?
  • Did you iterate at least once based on your reaction to the first batch?
  • Among all the ideas generated, were there any that you genuinely wouldn't have thought of yourself?

Reflection prompt for all four tasks

After completing whichever tasks you chose, take 5 minutes and write answers to these questions:

You don't need to share these answers with anyone. They're for you — to build the habit of deliberate reflection that turns practice into skill.