Using Machines Without Sounding Like One

You can tell within a sentence or two. A caption, an email, a website paragraph that reads a little too smooth, a little too balanced, using phrases nobody actually says out loud. It's not that the writing is bad. It's that it's nobody's writing. It belongs to a machine, and readers can feel the absence of a person even when they can't name why.

Machine-drafted text tends to be grammatically clean and emotionally flat at the same time. It hits every point evenly, uses the same sentence rhythm throughout, and reaches for the safest word instead of the true one. Human writing has bumps: an opinion stated too strongly, a sentence that trails off, a specific detail that only makes sense if you were actually there.

We use these tools constantly at Ariviz, for research, for first drafts, for restructuring a paragraph that's gone in circles. The problem was never the tool. The problem is publishing its first draft as your final voice, which is the writing equivalent of reading a stranger's cue cards out loud and calling it a speech.

Used well, a language tool is a fast first draft, a second opinion on structure, or a way to get past a blank page. It's excellent at organizing scattered thoughts into a logical order and terrible at knowing what actually matters to your specific customer, because it doesn't know your customer. It knows patterns from millions of other businesses, most of which are not yours.

The trap is using it to skip the thinking instead of speeding it up. If you ask a tool to write your About page from scratch with no input, you'll get a generic About page. If you dump your actual founding story, your actual complaints about your industry, and your actual customer feedback into it and ask for help organizing that into a page, you'll get something usable that still sounds like you, because the raw material was yours.

As more businesses lean on these tools for first drafts, the flat, average version of writing is showing up everywhere: emails, product pages, social captions, even reviews. That means a genuinely human sentence stands out more than it did two years ago, not less. Sounding like a real person isn't just a nice-to-have anymore, it's becoming one of the few remaining ways to be noticed at all.

There's also a trust cost to getting this wrong. Customers who sense they're reading something nobody actually wrote start to wonder what else about the business is running on autopilot. That suspicion is hard to shake once it sets in, and it tends to spread from the writing to the product itself, fairly or not.

When we write for a client, drafting tools handle the boring middle steps: organizing a messy interview transcript, checking a paragraph for redundancy, suggesting three headline directions to react to. The opinions, the specific stories, and the final sentence-level voice always come from a person who's actually talked to the client and knows how they'd phrase things. That split is what keeps the writing sounding like a business instead of a category.

If you're doing this in-house, the same split works without any special software. Someone on your team who actually knows the business should always be the last set of eyes before anything goes out, checking for one thing only: does this sound like us, or does it sound like anyone.

There's a version of this problem that shows up even for people who write carefully by hand: they read so much machine-drafted text now that its rhythms start leaking into their own writing unconsciously. Certain sentence patterns, certain transition phrases, certain overly balanced "on one hand, on the other hand" constructions, become familiar simply from exposure, and then show up uninvited in a founder's own emails. Noticing this is the first step to writing around it.

We keep a running list at Ariviz of phrases to flag on sight, not because they're technically wrong, but because they've become so common in machine-drafted text that they now read as a signal rather than a sentence. "In today's fast-paced world" is the most obvious offender, but there are dozens of smaller ones: "unlock your potential," "take it to the next level," "seamlessly integrate." None of these are grammatically bad. They're just worn smooth from overuse, and worn-smooth language is exactly what a reader's eye slides past without absorbing.

The upside of paying attention to this is bigger than avoiding a few stock phrases. Once you start editing for voice specifically, you get faster at spotting the difference between a sentence that's technically correct and one that's actually saying something. That skill transfers to everything else you write, from a sales email to a difficult conversation with a supplier, because it's really a skill about clarity and honesty, not about any particular tool.

None of this is an argument for writing everything slowly and by hand out of principle. Speed matters, deadlines are real, and a fast, decent draft beats a perfect one that never ships. The goal is simply to know which parts of the process you can hand off and which parts have to stay yours, so speed doesn't quietly cost you the thing that made your writing worth reading in the first place.