Services Work Insights Learn About Contact Book a free triage call →
Claude for Marketers

Keep it yours.

The lesson that makes the other five safe to scale: the table test, the editor pass, and the one-page policy that keeps your voice yours. AI drafts. You decide.

Lesson 06 of 07 · ~10 min · Free
The principle

AI drafts. You decide.

This is the non-negotiable that makes everything else in this series safe to scale. Nothing ships without a human pass, and the human is accountable. Not because AI output is bad, but because your judgement is the product your clients are paying for, and judgement can't be delegated.

The quick filter is the table test: would a smart friend say this sentence across a table? If a line couldn't survive being read aloud to a client, it doesn't ship. You'll be surprised how much AI-flavoured filler this one question removes.

Voice protection

Nobody decides to sound generic

It accrues one unedited draft at a time. Your voice is the asset AI is most likely to sand off, and the one clients actually pay for. Three habits protect it: keep a banned-words list, keep real examples of your best work in the Project, and run an editor pass on anything public before it goes.

Match the checking effort to the stakes. A brainstorm needs no verification. A statistic in a client deck needs a source. Anything legal, medical or financial needs a professional, not a chatbot. And some things shouldn't touch AI at all: apologies, sensitive client conversations, anything where the thinking is the deliverable and the struggle is the value.

Steal these

The two prompts that keep you honest

The editor pass

The last step before anything public ships.

Act as my editor. Check this copy against my voice: [three honest words].

Flag hype, jargon, corporate filler, and any sentence over 20 words. Apply the table test: would a smart friend say this across a table?

Score it out of 10, rewrite whatever fails, and tell me why it failed.

[paste your draft]

The claim check

Run before publishing anything with facts in it.

List every factual claim in this draft that I should verify before publishing: statistics, dates, names, prices, anything stated as fact. Rank them by how embarrassing they'd be if wrong.

[paste your draft]
Make it official

The one-page AI policy

If you work with anyone else, write this down. One page is enough. Tap to build your starting point.

AllowedDrafting, brainstorming, summarising, research starting points, formatting, editing your own work.
Draft-onlyAnything client-facing or public: AI drafts, a named human reviews and ships. The human's name goes on it.
NeverConfidential client data without permission, personal details, final legal or financial advice, apologies and sensitive conversations.
The sign-off ruleWho approves what before it ships, in one sentence. If everyone's responsible, nobody is.
The watchout

Silent drift.

The risk isn't one bad AI draft, it's a hundred unedited okay ones. Each sounds fine alone; together they sand your brand down to the average. The fix is the editing habit, not vigilance. Build the editor pass into the workflow and drift stops being possible.

Quick check

Three questions, then the good bit

No marks, no sign-up, just a check the lesson stuck. Tap your answer.

The table test asks:

How much verification does AI output need?

Brand voice drift happens through:

Lesson complete.

Judgement locked in. One lesson to go.

Next lesson →

Want your whole business working like this?

Brandling helps small businesses put AI to work without losing their voice. Senior help, right-sized.

Book a free triage →

Next lesson: Put it on autopilot →

Book a free call Email me