Write With AI Without Sounding Like Everyone Else

Posted September 22nd, 2026 under AI

Small business owner writing notes by a window, planning content before she decides how to write with AI without losing her own voice

Ninety-seven percent of content marketers plan to use AI for content marketing in 2026, and only seven percent of consumers say AI-generated content makes them trust a brand more. Everyone is reaching for the same tool. When you write with AI and publish the first draft untouched, you are not saving time — you are borrowing your competitor’s voice, because there is a good chance they used the exact same prompt.

Why AI writing sounds the same everywhere

Every large language model is trained on similar data and nudged toward the same “safe,” inoffensive register, so they all reach for the same words. Researchers tracking word frequency across published writing found that “delve” now appears roughly 25 times more often than it did before ChatGPT existed. Opening lines like “in today’s fast-paced world” and “in the digital age” are reliable tells, and so are rule-of-three lists (“efficient, scalable and reliable”) and inflated verbs — utilize instead of use, leverage instead of help. None of it is wrong exactly. It is just recognizable, and recognizable is the opposite of what a small business wants from its own website, proposal or email.

What sounding generic actually costs you

When you write with AI for something a real customer will read, the data on how people react is blunt. A Klaviyo and Datalily survey of 8,000 consumers across eight countries found only 7% say AI-generated marketing content increases their trust in a brand, while 31% say it decreases trust outright. Separate research from Emplifi found that 52% of consumers say they would stop buying from a brand after content that felt inauthentic. AI detectors have also gotten good enough that guessing wrong is risky on its own — tools built for the job now claim upward of 99% accuracy at flagging machine-written text. Writing that reads as generic used to be a style complaint. Now it is a trust problem, and trust is the one thing a small studio or a small shop cannot outsource to a language model. We covered a related side of this when we looked at what AI can and cannot do in a real design workflow — the same line holds for words as it does for images.

Close-up of hands writing edits on a notepad, marking up a first draft before it goes live

What this looks like for a small business in Aruba

Aruba raises the stakes rather than lowering them. Most small businesses here are known by name in their own neighborhood, and a website or a WhatsApp reply written in flattened, generic English reads like a stranger talking, not the person a customer met at the counter last week. The island runs across English, Spanish, Dutch and Papiamento in daily life, and AI writing tools trained overwhelmingly on English source material tend to flatten every language they touch into the same bland register. A beach bar in Palm Beach and a dive shop in Savaneta should never sound alike online, no matter which tool drafted the first pass.

Five ways to write with AI without losing your voice

  • Feed it your own words first. Paste three or four of your own past posts, emails or reviews into the chat before you ask for anything new, so the draft has your sentence rhythm to copy instead of the model’s default one.
  • Ban the tell words. Keep a short list — delve, unlock, elevate, seamless, “in today’s” — and strip every one out in the edit pass, even the ones that read smoothly.
  • Ask for options, not a finished page. A prompt like “give me three different ways to open this” produces raw material you choose from and rewrite, instead of a paragraph you’re tempted to publish untouched.
  • Write the opening and the ending yourself. These are where AI drafts get vaguest and most falsely upbeat. A specific, human opening and closing line does more for trust than anything the model adds in between.
  • Read it out loud before it goes live. If it doesn’t sound like something you would actually say to a customer standing in front of you, it needs another pass.
Two colleagues reviewing a printed page of business copy together at a desk, candid conversation

A workflow that keeps you as the author

Outline the piece yourself first — three or four bullet points on what you actually want to say and why a customer should care. Let AI expand that outline into a rough draft, then edit hard: cut every stock phrase, add one concrete detail only your business would know (a real number, a real customer situation, a real local reference), and rewrite the opening and closing paragraphs by hand. Treat the AI draft the way you would treat a first draft from a junior writer on your team — useful raw material, never the finished piece. Whether you write with AI once a week or every single day, that one habit is what keeps the words sounding like you instead of like the model.

Do this before your next AI-assisted post

  • Search your last three posts or emails for “delve,” “unlock,” “elevate” and “in today’s,” and cut every hit you find.
  • Read your homepage intro out loud and ask whether it sounds like you or like a brochure written about you.
  • Add one fact, number or local detail to your last AI-assisted draft that only your business could have written.
  • Set a house rule: no AI draft publishes without a human edit pass and a spoken read-through first.
  • Check your welcome email and About page first — these get read closely, and generic language there costs the most trust.