The proof is really in the pudding, isn't it? I don't see a wave of successful vibe-coded startups in the market yet. That's kind of the benchmark for whether this stuff actually does in practice what the AI-hypemen claim it can.
Rather the opposite. A vibe-coded startup cannot survive if it can be trivially duplicated. The proof will be in observing the inverse phenomenon; (pure) software companies disappearing.
Making someone’s agents 20% better, cheaper or faster will be a measurable and easy sales goal.