Great AI conversations start with great instructions. This lesson breaks down structure, tone, guardrails, and the small edits that lift reply rates.
Every instruction set benefits from four sections: identity, goal, rules, and hand-off.
The AI is agreeable by default. Guardrails come from explicit 'do not' rules: don't quote rates, don't guess about programs you don't offer, don't push past a soft no.
Read five real conversations a week. Every awkward reply is a missing rule. Add it and move on.
Long enough to cover your rules, no longer. A tight page of bullet-style rules typically outperforms a long essay.
Yes — treat them like a shared playbook and version them.
Your AI is only as good as your instructions. This lesson is the difference between an assistant that sounds like a chatbot and one that sounds like you.