A Month of Social Posts That Don't Sound Like a Robot
12 min read
Social media is where AI writing goes to be spotted. You've seen the posts — the rocket emoji, the "We're SO excited to announce," the paragraph that says nothing in the friendliest possible way. Your customers scroll past those without registering them, which means the time spent posting them was worse than wasted.
The antidote isn't posting less. It's giving the AI two things it never gets: your voice document, and raw material from your actual week.
Raw material beats imagination
The single biggest upgrade in AI social content is this: stop asking it to invent and start feeding it reality. Before you prompt, spend five minutes jotting what actually happened lately. A pizza shop's list might be: tried a new dough supplier, a regular ordered the same thing for the 100th time, hiring a weekend driver, the oven repair saga finally ended.
That list is a month of posts. Generic AI content happens when the AI has nothing real to work with — so give it something real.
The batch prompt
Here is my brand voice document: [paste]
Here are notes on what's happening at my business this month:
[paste your list]
Write 12 social posts I can spread across the month:
- 4 based directly on my notes (the real stories)
- 3 useful tips a customer would appreciate, drawn from what a
business like mine knows
- 2 behind-the-scenes or team posts
- 2 that mention a service or product, low-key, no hard sell
- 1 that asks the audience a genuine question
Keep each under 80 words. Vary the openings — no two posts start
the same way. No hashtag walls: 0-2 hashtags where natural.
Where a photo would help, add a one-line note in [brackets]
about what to shoot.The mix matters. A feed that's all promotion gets muted; the 4-3-2-2-1 spread keeps you interesting between the asks. And the [photo note] trick quietly solves the other half of social media — you now have a shot list, and a phone photo of your actual shop beats a stock image every time.
Editing pass: hunt the AI-isms
Read the batch and cut ruthlessly. Delete any post you can imagine a competitor posting word-for-word — if it's not specifically yours, it's filler. Kill exclamation marks beyond the rare earned one. Replace any "We're excited/thrilled/proud" opening with the actual news. Then fix the details the AI got wrong about your own stories; it was working from three-word notes, and you were there.
Fifteen minutes of editing on a 12-post batch, and you'll still come out far ahead of writing them one guilty Sunday at a time.
Platform fit, the cheap way
Don't regenerate from scratch per platform — convert:
Take post #3 and adapt it: an Instagram caption (looser, line
breaks welcome), a Facebook version (a sentence more context),
and a LinkedIn version only if it makes sense for a business
audience — if it doesn't, say so instead of forcing it.That last clause is worth keeping. An AI told to always produce a LinkedIn version will produce one whether or not it should — inviting it to say "this one doesn't fit" is how you get an assistant instead of a yes-machine.
Try it now
Write your five-minute reality list — what actually happened at your business this month. Run the batch prompt with your voice document. Post the single best result this week, with a real photo. One genuinely specific post outperforms seven generic ones, and now you know how to make twelve.