Question
You're the team's expert on your company's custom event-sourcing framework — it has unusual rules about idempotency keys and event ordering that aren't documented publicly. You ask an agent to add a new projection, and it writes confident, fluent code that follows mainstream event-sourcing conventions but violates two of your framework's invariants. The code compiles and passes the agent's tests. How do you handle working with AI in a domain where it will be confidently wrong, and how do you keep this from shipping?
Treat the AI’s output as a draft to verify, not an answer to trust. Name the specific flaw and the input that triggers it, say how you’d catch it — tests, edge cases, reading critically — and how you’d re-prompt or decompose to get it right.
Vibe coding: describe the solution in plain language (or narrate it) and the coach grades your approach. Generating runnable code from your description is coming next.