The AI Content Standards Field Guide
A method for setting content standards that hold consistency across every generated output. Seven categories, a worksheet that becomes your product's standards doc, and a Claude skill that checks your copy against all of it.
Your product ships AI features and nobody has decided what good copy looks like across them. The empty state reads warm, the error clinical, the confirmation chatty, because each was generated against a different implicit prompt. The problem only shows up after something ships.
This guide gives you the seven categories where AI copy needs a standard, the line that holds in each, and where your product is free to vary by its own risk. The worksheet turns your decisions into a standards doc in the words engineering pastes straight into a prompt. Built to use mid-task, for an initial build, or for a full system overhaul.
What's inside
The Field Guide (20-page PDF). The framework for a standard that holds, all seven categories with a worked example each, and one complete filled example for a B2B SaaS assistant.
The worksheet (PDF + Notion template). One page per category. You fill four fields; the Instruction field hands straight to engineering.
The Claude skill. Give it a piece of your AI copy and it names what's wrong, rewrites it to standard, and hands you the rule that prevents the repeat.
Who it's for
You shipped an AI feature and no one decided what it says when it's unsure, when it fails, or when it hands off to a human.
A cheerful default landed on a user in a hard moment and you need a rule so it doesn't happen twice.
You're building the standards your whole team will write AI features against.
