Question
Two engineers report wildly different experiences with the same AI agent. Engineer A used it to migrate a 2,000-line legacy module from JavaScript to TypeScript and says it saved a week. Engineer B used it to design a distributed rate-limiter and says it 'felt like a week of progress in a day' but the design fell apart in review. As a tech lead, explain the underlying difference: when does an AI agent genuinely speed you up versus merely manufacture a feeling of progress?
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.