A feature that would proactively check whether the agent is abusing the third-party AI while replying to reviews, as currently it can be done only by manually exporting their answers and checking them with Gemini.