How to use AI for Social Content without the Slop

You know the stuff. Perfectly grammatical. Politely enthusiastic. Absolutely nobody’s favourite. The kind of post that sounds like it was written by a committee of accountants who all share one LinkedIn password, then sprinkled with AI buzzwords like elevate and deep dive.

That’s AI slop: content that is technically fine, but emotionally empty. It gets scrolled past faster than a terms and conditions page.

The good news is AI does not automatically produce slop. People do. Usually by asking for “10 engaging posts about my business” and then publishing the first thing that comes out. AI is a power tool. If you use it like a butter knife, don’t be surprised when dinner is a bit tragic.

This post is about using AI to create quality social media content that feels human, specific, and worth reading, while saving you loads of time.

First, define what quality looks like for your audience

Before you craft a prompt, decide what quality means for your audience and potential customers. Here are four quick tests.

1) The “could anyone post this” test

If a competitor could copy-paste it, change the logo, and it still makes sense, it’s generic.

2) The “smell test”

If it reads like a brochure, it will perform like a brochure.

3) The “so what” test

If the reader cannot immediately tell why it matters to them, they’ll keep scrolling.

4) The “right person” test

If your ideal audience or potential customer read it and would not think, this is for me, it’s probably trying to please everyone. Pleasing everyone is how you end up with content that excites nobody.

So your goal is not “use AI to write posts”. Your goal is “use AI to produce better raw material, faster, so you can publish content with a point of view”.

The real secret

AI is best as a thinking partner, not a typing machine.

Most people use AI like a vending machine: insert prompt, receive post. That’s where the sludge comes from.

Instead, use AI for the parts humans tend to rush:

• finding stronger angles

• clarifying the audience’s real problem

• generating examples

• discovering alternative viewpoints

• tightening structure

• adapting one idea across multiple formats

Then you, the human, do the high-value bits:

• choose the opinion

• add the story

• sanity-check claims

• inject lived experience and brand tone

The aim is a hybrid workflow where AI does the heavy lifting and you do the taste-testing.

Get the foundations right first

Once the setup is done, the weekly workflow becomes almost boringly repeatable, which is exactly what you want.

If you want AI to sound like you, you need to give it something to work with. Otherwise it will default to generic, overpolite internet mush.

Set custom instructions

Write a short set of custom instructions that act like a brief you would give a human copywriter. Keep it practical:

• who you help, and who you are trying to attract

• the outcomes you want from social content, for example leads, event sign-ups, enquiries, retention

• your non-negotiables, for example UK English, no clichés, no hype, no vague advice

• your content boundaries, for example what you do not post about, like no vague motivation posts and no hypey ‘revolution’ language, plus avoiding those AI buzzwords

The clearer you are on what you will not say, the easier it is for AI to stay on-brand and avoid that generic, copy-paste vibe.

Use a Project or Notebook, not random chats

If you are serious about consistency, set up a dedicated Project in ChatGPT or a Notebook in 365 Copilot (licensed version). This is where you keep:

• brand guidelines, including voice and tone rules

• example posts you are proud of

• audience notes, pain points, typical objections

• product and service info, offers, pricing and FAQs

The aim is simple: the AI starts every session already understanding the context, instead of you re-briefing it every time.

Be specific about voice and tone

Most brand guidelines focus on colours and logos. Useful, but not enough. Add a mini voice guide so the AI can mimic your style:

• sentence rhythm, for example short punchy lines with the occasional longer reflective line

• language preferences, for example plain English, fewer buzzwords, more concrete examples

• personality cues, for example friendly, direct, lightly witty, but never snarky

• platform differences, for example LinkedIn more reflective, Instagram more direct

If you have posts that feel spot on, drop in three to five examples. AI learns patterns quickly when you show it what good looks like.

Use memory to build momentum

If your AI tool supports memory, use it deliberately. Ask it to remember:

• topics and themes you want to revisit

• stories, case studies, and examples you can reuse

• your preferred content formats, for example myth-busting posts, checklists, short story-led lessons

• recurring content pillars, for example AI productivity, social media best practice, practical training tips

That way, next time you say, give me a post idea for this week, it can suggest something aligned with what you have already said you care about.

A practical workflow you can reuse every week

Here’s a simple process that works for solo business owners, marketers, and teams.

Step 1: Start with one real moment, not a topic

Slop begins with vague prompts. Better content starts with something concrete:

• a customer question you received this week

• a mistake you see clients making

• a result you achieved

• a common myth in your industry

• a strong opinion you can defend

Write one sentence in plain English:

Today I want to talk about X because people keep getting Y wrong, and it costs them Z.

Then feed that to AI.

Step 2: Ask AI for angles, not finished posts

Try prompts like these (edit the brackets):

Here’s my rough idea: [paste]. Suggest 10 angles that would be interesting to [audience]. For each angle, include the likely objection and a punchy opening line.

Turn this idea into 5 different viewpoints: contrarian, practical, story-led, myth-busting, and ‘here’s what I’d do if I started again’.

Pick one angle that feels like you, not one that sounds like a marketing handbook.

Step 3: Build the post around a spine

Great posts usually have a clear spine. Choose one structure and stick to it:

• Hook → Problem → Fix → Example → Call to action

• Myth → Truth → Proof → How to apply

• Story → Lesson → Framework → Next step

• Mistake → Why it happens → Better approach → Checklist

Ask AI to outline before drafting:

Outline a post using Hook-Problem-Fix-Example-CTA. Keep the hook under 15 words. Make the example specific to [your context].

Step 4: Generate specifics, because specifics kill slop

Generic content is vague by default. So ask for specifics deliberately:

Give me 6 realistic examples from [industry] that illustrate this point.

Suggest 10 ‘wrong but common’ phrases people use, and rewrite them into something clearer.

List the top 5 questions my audience will ask after reading this, then add one line answering each.

If you only do one thing to avoid slop, do this: force detail into the draft.

Step 5: Do a human pass before you publish

This is where you earn the win.

Add at least two of the following:

• a personal observation, even small

• a clear opinion, with reasoning

• a tiny story, what happened and what you learned

• a named tool or method you actually use

• one hard-earned warning, what to avoid

Then edit the voice to avoid the obvious AI tells:

• overpolite filler, In today’s fast-paced world

• bland hype, unlock the power

• generic advice, be consistent, with no how

Ask AI to help you cut it:

Rewrite this in UK English with a confident, friendly tone. Remove filler and clichés. Keep it punchy. Make it sound like someone with real experience, not a brochure.

Step 6: Repurpose properly, not lazily

Repurposing is where AI shines, if you give it guardrails:

Turn this post into a 30-second script, 5 carousel slide headlines, and 3 short replies I can use in the comments. Keep the point of view consistent.

Create two versions: one for LinkedIn (slightly more reflective), one for Instagram (more direct). Keep the examples.

You keep the idea. AI adapts the format.

The slop prevention checklist

Before posting, check these:

• does the first line earn attention, or just announce the topic?

• is there at least one specific example that feels real?

• can a competitor post this without changing much?

• have you included a clear opinion or a useful distinction?

• did you remove marketing fog phrases?

• would you stop and read this if it wasn’t yours?

If you can tick most of these, you’re not publishing AI slop. You’re publishing content you used AI to craft faster.

Three prompt templates you can steal

1) The angle generator

Audience: [who]. Goal: [what you want]. Topic: [what it’s about]. Give me 10 angles that are not generic. Include a strong hook for each and one practical takeaway.

2) The specificity injector

Here’s my draft: [paste]. Add specificity: examples, numbers where appropriate, and common mistakes. Do not invent case studies. If something needs evidence, flag it.

3) The brand voice editor

Rewrite in a clear, friendly, confident UK tone. Keep sentences varied. Remove clichés and filler. Keep it grounded and practical.

Final thought

AI won’t save you from boring content. But it will save you time, and time is what lets you think properly again.

Use AI to do the grunt work, then put your judgement, stories, and standards on top.

That’s not slop. That’s content that works for you and your audience.

Want help setting this up properly?

If you’d like a hand setting up your custom instructions, Projects or Notebooks, plus a repeatable content workflow your team will actually use, book a session and we’ll get it sorted.