Lesson 2 · Find a problem worth solving: from a hunch to one real conversation

Where you are

In Lesson 1 you listed three problems. Today you choose one, describe it clearly, and check it with a real person. You are not building anything yet. Lots of products fail because nobody checked the problem first, so this hour matters.

What you will make

A problem statement: two or three sentences that say who has the problem, when it happens, what goes wrong and what it costs them. You also make five good questions and use them in one short, real conversation.

For the fictional café: "The volunteer coordinator (who), every Sunday evening when she plans the week (when), finds empty slots on the rota and doesn't know who is free (what goes wrong), so she spends two hours phoning and texting, and some shifts still go unfilled (what it costs)."

Outcomes

By the end of the tutorial you can:

1. Choose one problem to work on.
2. Write a problem statement that does not mention a solution.
3. Write questions about what people did, not what they might do.
4. Practise a conversation with the AI, knowing it is only practice.
5. Ask one real person about the problem and note what they said.

How to use the prompts

• Copy each prompt as written, paste it into your AI assistant and press send.
• Replace anything in [square brackets] with your own words, and delete the brackets.
• Every prompt works in ChatGPT, Claude or Gemini, on a free account.
• After each prompt, read the Check and answer it honestly before moving on.
• Never paste personal information: no names, addresses, phone numbers, health details, money details or anything about the people you help or work with.

Tutorial

1 · Pick one problem 5 min

Here are three problems I have noticed: [problem 1], [problem 2], [problem 3]. For each one, ask me one question that helps me see how often it happens and how much it matters. Then help me choose the one to work on first.

Check. Did you choose one problem that happens often, matters to someone, and that you know a real person who has? If you can't think of anyone who has it, did you consider choosing a different one?

2 · Write your problem statement 8 min

Help me write a problem statement for this problem: [your problem]. Use this shape: who has the problem, when it happens, what goes wrong, and what it costs them (time, money, stress or a missed chance). Do not mention any app, website or solution. Keep it to three sentences. Ask me two questions first if you need more detail.

Check. Did you read the statement aloud? Does it say who, when, what goes wrong and what it costs? Did you remove any app, website or "AI" from it? Copy the final version into your notebook.

3 · Write five good questions 7 min

Write five questions I can ask a real person to find out whether they have this problem: [your problem statement]. Ask about the last time it happened and what they actually did, not what they would do in the future. No question should be answerable with a polite "yes". Do not mention my idea.

Check. Did you check each question by asking yourself "Could they just say yes to be kind?" If so, did you change it to start "Tell me about the last time you..."? For the café, "Would you use a rota app?" becomes "Tell me about the last time a shift was empty. What did you do?"

4 · Rehearse with the AI 8 min

Practice only. Nothing the AI says here counts as evidence, because it is not a real person.

Practice only. Pretend to be [the kind of person who has the problem], a busy person who is polite but a little doubtful. I will ask you my questions one at a time. Give short, realistic answers, and don't make the problem sound worse than it is. When I type "STOP", tell me which of my questions worked well and which one I should change.

Check. Did you change at least one question because of the feedback? Did you keep the AI's answers out of your notes about the problem? The AI only helped you practise.

5 · Ask one real person 7 min

I am about to ask a real person about this problem, for about five minutes, [in person / by phone / by message]. Write me a friendly two-sentence opening that says why I am asking, that there are no right or wrong answers, and that I won't write down their name. Then give me a simple notes page with two columns: "What they did" and "What they said".

Now talk to your person, using your best three questions. If they are not available right now, book a time before Lesson 3. Listen more than you talk, and don't describe your idea.

Check. Did you talk to a real person, not the AI? Did you write "Person 1" and their role (for example "café volunteer"), not their name? Did you note at least one thing they actually did, and one thing that surprised you? If they don't have the problem, well done: you found that out before building anything. Pick another problem or another person.

Facilitator notes

• Many learners pick a problem that is too big ("loneliness", "climate change"). Encourage a small, everyday moment: who, when, and what goes wrong.
• Step 5 works well in pairs within the group: each learner interviews someone from another table who might have the problem. Say clearly that a classmate is a stand-in if they don't really have it.
• Watch out for problem statements that are really solutions ("People need an app for..."). Ask "What happens today, without the app?"
• If anyone's problem touches health, money, legal matters or safeguarding, remind them to keep details general and to talk to a qualified person about the real situation.
• Some learners will feel shy about asking a real person. Offer to be the first person they ask, or let them send their questions as a message after the session.

© 2026 MVP Studios Ltd. The Build with AI: First Steps course materials are free to use, print, translate and adapt for non-commercial training under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 licence: credit MVP Studios Ltd and share adaptations on the same terms. MVPAI™ and the MVP Studios name are trademarks and are not licensed. Course terms