AI Workflow

Because multiple language models I use often produce meaningless filler, user-pleasing but misleading phrasing, and sometimes even distorted facts, I designed a behavior protocol. I use this protocol in two situations: Before each conversation. Mid-conversation when the model shows the above issues. Over about a week of testing with Gemini, I continuously improved the protocol’s completeness, rigor, and practical effectiveness, including self-check and self-judgment logic.