No AI Language: Rules for Writing That Sounds Human

AI writes faster than any human. It also writes in a way most readers can now spot in a sentence or two. These rules exist to keep the speed and lose the tells.

Use this as a working style guide for anything written with AI assistance: client emails, landing pages, decks, captions, briefs. The pattern is simple. Each item names what to avoid and what to use instead. When something on the left shows up in a draft, rewrite.

Formatting

Do not use em dashes. Use plain dashes or commas.

Do not use ellipses. Use periods and short, direct sentences.

Do not overuse parentheses. Keep punctuation simple and minimal.

Do not auto-format into bullet lists. Use bullets only when explicitly requested.

Do not insert markdown headers in chat responses. Write in plain text unless headers are asked for.

Do not bold or italicize unless requested. Use minimal formatting only for clarity.

Style

Do not use "it isn't this, it's that" constructions. Answer with straightforward phrasing.

Do not end with "in conclusion" or "to summarize." End cleanly without a wrap-up.

Do not start with "Certainly" or "Of course." Start with the answer.

Do not add fluff like "here's the thing" or "let's dive in." Be concise and direct.

Do not give long wind-ups. Answer immediately, then expand if needed.

Do not include "as an AI language model" disclaimers. Stay in character and answer.

Do not fake enthusiasm with "that's a fantastic question." Keep a professional, natural tone.

Do not overuse "however," "moreover," or "furthermore." Use simple transitions like "but" and "and."

Do not frame with "imagine if" unless requested. Stick to direct explanations.

Tone

Do not hedge with "may," "might," or "possibly." Be confident and direct.

Do not pile on disclaimers. Include a disclaimer only when it is essential.

Do not explain both sides when neither was requested. Keep it informative and practical.

Do not take a preachy or moralizing tone. Stay straightforward and useful.

Do not write like a high school essay. Use conversational, real-world phrasing.

Repetition

Do not restate the user's question. Answer directly without repeating the prompt.

Do not repeat phrases inside the same paragraph. Keep responses lean.

Do not explain the same point in three different ways. State it once, clearly.

Do not recycle stock phrases across answers. Use varied, original language.

Sentence-Level Habits

Do not lean on colons and semicolons. Use commas and periods.

Do not write rambling sentences. Favor short, punchy ones.

Do not use redundancies like "for example, such as." Cut one of them.

Do not default to "if you want to" framing. Say "do this."

Do not start sentences with "so" unless the moment is truly conversational.

Structure

Do not end with a list when the request was a narrative. Match structure to the ask.

Do not break things into numbered lists by default. Keep responses linear.

Do not wrap up with "ultimately," "overall," or "in summary." End simply.

Do not over-explain the obvious. Prioritize clarity and brevity.

Banned Words and Phrases

These are automatic disqualifiers. If they appear in a draft, rewrite the sentence.

Corporate cliches

Synergy, leverage, ecosystem, game-changer, disruptive, impactful, utilize, optimize (outside a technical context).

AI and coach-speak

"Unlock your potential," "manifest," "I'm so passionate about," "in today's fast-paced world," "journey," "unleash," "empower."

Minimizing words

Just, simply, basically.

Hustle culture

Hustling, grinding, no excuses, rise and grind.

Filler and hedge words

Genuinely, honestly, straightforward, perhaps, "it could be argued."

Shame and urgency tactics

"Summer is coming, are you ready," "don't miss out," "limited time offer," "act now," "burn off that weekend."

How to Use This

Treat the list as a filter, not a script. Draft fast. Then read the draft once with this guide open. Pull the dashes. Cut the wind-ups. Replace the banned words. The rewrite usually takes less time than the first draft.

Writing that sounds human is not a style choice. It is the price of being read.