Start with a problem worth solving
Choose one customer and one expensive, repeated problem.
01Start with the work, not the tool
A business earns money by solving a problem someone cares about. AI can help you deliver that solution, but “I use AI” is not a customer benefit. Choose a group you can actually reach: independent trainers, local service businesses, or small online shops. Describe a repeated task that costs them time, loses enquiries, or delays delivery.
02Look for evidence
Interview five people in that group. Ask what happened the last time the problem occurred, how they handle it now, and what the delay costs. Avoid asking whether they like your idea: polite encouragement is weak evidence. Write down their exact words and any existing spending. AI can organize your notes; it cannot replace the interviews.
03Make a small bet
Compare three problems on frequency, urgency, access to customers, and your ability to deliver. Score each 1–5 and explain each score. Choose the strongest evidence, not just the highest total. For example: help a local trainer turn one approved weekly tip into three social posts, with the trainer reviewing every post. Start with a paid pilot before building an elaborate product.
Your assignment
Write: I help [specific customer] solve [repeated problem] by delivering [concrete result]. List five people to interview and three questions about what they do today.
Your starting prompt
Act as a skeptical business researcher. My audience is [audience]. Their problem is [problem]. Here are my interview notes: [notes]. Separate observed facts from assumptions. Identify contradictions, suggest five non-leading follow-up questions, and propose one small test. Do not invent customer evidence.
Replace the brackets with your context. Review the result before using it.Members can save notes and track completed lessons. You can try the exercise here before joining.