Can AI Write Social Media Posts That Sound Like You?
AI can write in your voice, but only if you brief it properly first — the default setting is the statistically ordinary version of every business in your trade. Here's what actually separates output you'd publish from output your regulars can spot a mile off.
Dave Smith
# Can AI Write Social Media Posts That Sound Like You?
Short answer: yes, but not on its own, and not straight out of the box.
That's a less exciting answer than either side of this argument usually gives you. One camp says AI writing is indistinguishable from human writing now and you'd be daft not to use it. The other says you can spot it a mile off and customers will hold it against you. Both are partly right, and the bit that decides which one you experience is what you feed the thing before you ask it for anything.
Here's the thing about AI and tone of voice. The model isn't guessing at your personality from thin air. It's producing the most statistically ordinary version of whatever you asked for. Ask it for "a social media post about our new opening hours" and you'll get a post that sounds like every business that has ever posted about opening hours — rocket emojis, "we're excited to announce", an unearned exclamation mark. That's not the AI failing. That's the AI doing exactly what you asked, which was nothing in particular.
What "sounds like you" actually means
Most people can tell when something doesn't sound like them, but struggle to say why. It's worth pinning down, because you can't brief a tool on a feeling.
Your voice online is mostly four things. Formality — do you write "we are pleased to confirm" or "all sorted, see you Tuesday"? Warmth — are you matey or professional-friendly? Directness — do you get to the point in one line or build up to it? And humour — dry, none, or full-on daft?
Then there's vocabulary. Every trade has words the people inside it use and words the marketing brochure uses. A joiner says "second fix". A brochure says "finishing touches". If your posts use the brochure words, your regulars will notice something's off even if they couldn't name it.
The good news is that all of this is describable. It's not magic. It's a set of dials, and a decent AI setup lets you set them rather than guess.
Where AI genuinely gets it wrong
Being honest about the failure modes matters more than the sales pitch, so here are the ones that actually catch people out.
The enthusiasm problem. Untuned AI writes everything at an eight out of ten excitement level. Reordering stock is not an eight. If your posts sound permanently thrilled, they read as insincere within about a fortnight.
Specificity that isn't real. Ask an AI for a post about your bakery and it may cheerfully mention your "signature sourdough" whether or not you make sourdough. It's filling a gap with something plausible. Anything factual — prices, timings, product names, guarantees — needs checking by someone who knows the business, every single time. This is the one that can genuinely damage you.
Tics. Certain constructions turn up constantly: "In today's fast-paced world", "It's not just X, it's Y", rhetorical questions stacked three deep. None are wrong exactly. They're just visibly machine-shaped once you've noticed them, and your customers have started noticing them.
The flatness. Even well-tuned AI writes competently rather than memorably. It won't produce the odd, specific detail that makes a post land — the customer who brings her dog in every Thursday, the van that won't start in the cold. You have those. The AI doesn't.
What actually works
The setups that produce publishable output all do the same three things.
Feed it real material about your business first. Not a paragraph of description — actual source material. Your website copy, your FAQs, how you answer the phone, three posts you were happy with. The more genuine input it has, the less it has to invent.
Set the tone deliberately rather than accepting the default. If your tool lets you adjust formality, warmth and the rest, adjust them. The default is the middle of every scale, and the middle of every scale is where every other business's content lives too.
Edit for one thing, not everything. You don't need to rewrite AI output line by line — that defeats the purpose. Do one pass looking for anything factually wrong, and add one specific detail only you would know. That's typically thirty seconds of work and it's the difference between a post that could belong to anyone and one that clearly belongs to you.
So should you use it?
Depends what you're comparing it to.
If you're comparing AI-written posts to the ones a talented copywriter produces after a proper discovery session, no, they're not the same. They won't be for a while.
But that's rarely the actual choice. For most small businesses the real comparison is between AI-assisted posts and nothing at all — because posting has been on the to-do list since March and it's July now. Against that, a well-briefed AI post that's ninety per cent right and takes a minute to check is enormously better than a page that's gone quiet since spring. Customers don't award points for the posts you meant to write.
The businesses that get value from this treat AI as a very fast, slightly generic first draft written by someone who's read your website but never met you. You wouldn't publish that person's work unread. You also wouldn't turn down the help.
That's roughly the principle Aunty Social is built on — it reads your website to learn what you actually do, then works from a tone profile you set across five sliding scales, so the output starts in your register rather than the internet's. It's not pretending the last mile doesn't exist. It just means the last mile is a quick read-through instead of a blank page on a Sunday night.
Start with tone, check the facts, add the detail only you know. Do that and the honest answer to the question in the title is yes.