Imagine you’d have an AI agent fine tuned to generate end-to-end tests. What would be your preferred way to tell the AI agent what test do you want? A. Define your own test cases step for step in natural language B. Choose from a curated list of test case descriptions C. Use a test recorder D. Anything else?

For disclosure - I’m doing market research for my company. We’ve developed an AI agent that generates Playwright tests automatically. We are refining the test delivery and your answers would really help. We ask devs who've worked with testing tools (code or no-code) about their preferred way to tell ‘a system that does the writing’ what do you want to test.