This is a product note in progress. The question is not simply where AI can be added, but which problems are worth solving differently with AI and how model capabilities can become part of a complete product loop.
Questions I want to answer
- How do users complete the task today, and where does it demand the most judgment?
- Is the problem frequent and important, and can it tolerate some uncertainty?
- Does AI generate, compress information, support decisions, or take action?
- How can users understand, correct, and verify the result?
- If the AI layer disappears, what distinctive product value remains?
What I will add next
I plan to turn these questions into a practical scenario-screening worksheet and compare capability-led demos with products that earn a durable place in a real workflow.
This note will continue to evolve alongside my product practice.