01 Start with the customer’s task
Write down who uses the product, who chooses it and who would pay for it. Those can be different people. Describe the current workflow in enough detail that someone outside your industry can follow it.
Prepare answers to: What triggers the need? What does the customer use today? What have you learned from actual conversations? Keep interview findings separate from your own assumptions.
02 Show what your AI can demonstrate
Prepare one complete example with an input, output and review step. Label a recorded demonstration clearly. Use sample data that you have permission to show.
Ask yourself: What did we evaluate? What was the comparison? How did we select the examples? Where does the system fail? A useful answer includes the test conditions and the limitations, alongside the result.
03 Explain dependencies and delivery
List the outside models, data sources and integrations the product needs. Explain which parts your team built and what would need attention if a provider changed.
Be ready to discuss how you track model usage, hosting and human review. Separate measured delivery costs from estimates. Describe your handling of customer data using the process you actually follow.
04 Keep the meeting useful
Prepare a short evidence list, with an owner and a date for each item. Mark unanswered questions as open and write down what would resolve them. Do not turn a planned feature into a present capability.
Finish by agreeing on the next piece of information to share. Record questions from the meeting so your next update addresses what the person actually asked.