Try this. Open ChatGPT and ask for a blog post on why small businesses need managed IT services.

You'll get a perfectly decent post. It will mention downtime, security threats, predictable costs, and the value of a proactive partner. It will be well organized and free of typos.

Now imagine every other MSP in your market asking the same question this week. They're getting roughly the same post.

That's the quiet problem with AI content right now. It isn't bad. It's average. And average doesn't win you clients.

Sounding Alike Twice Over

MSPs had a sameness problem long before AI showed up.

I've reviewed nearly 200 MSP websites, and most of them reach for the same words: proactive, comprehensive, trusted partner. A lot of MSP content comes from the same few marketing programs, which hand the same blog posts and newsletters to hundreds of firms. Swap the logo and the post works for anyone.

Now many of those same MSPs are using AI to keep up. The tool is drawing on the same general pool of MSP content that was already interchangeable. So the output sounds like the templates, which sounded like each other.

That's sounding alike twice over. And it happens without anyone noticing, because every individual piece looks fine.

AI Isn't Going Anywhere, and It Shouldn't

I'm not going to tell you to stop using AI. For most MSPs, that isn't realistic.

You don't have a full-time writer. The people who know the most about your business are busy running it. Content still has to go out every week, and AI is the reason a lot of MSPs can publish at all.

So the question isn't whether to use AI. It's how to use it without disappearing into the average. The answer has less to do with the tool than with what you give it.

Why the Output Goes Generic

An AI tool doesn't know anything about your MSP unless you tell it. It doesn't know about the client you kept through a ransomware scare, or why you stopped taking accounts under 20 seats, or how your service manager explains a backup failure to a nervous office manager.

When it has nothing specific to work from, it fills the gap with what's most common. For MSPs, what's most common is "proactive," "comprehensive," and "peace of mind."

That's why rewriting the prompt rarely fixes it. You can ask for a more conversational tone or a punchier headline, and you'll get a more conversational, punchier version of the same generic post. The problem isn't how you ask. It's what the tool has to work with.

Input One: Material Only You Have

The freshest content an MSP can publish comes from inside the business. It's the stuff a competitor can't copy because they didn't live it.

Think of the onboarding process you built after one bad client transition years ago, or the rule you made about which clients you'll take. Or the way your ops lead explains a problem on a call, in plain language that lands better than anything on your website. I wrote about where this kind of evidence hides most often in The Three Places Operational Evidence Hides.

None of that is on the internet, so no AI tool can produce it on its own. But once you hand it over, the tool can do a lot with it. A ten-minute voice memo from your ops lead about a recent client problem gives AI more to work with than any prompt you could write.

If you're not sure where to start, your best material comes from operational patterns. These are the places it turns up most often, each with a question worth asking your team:

The job, then, is to get that material out of people's heads on a regular basis. Take one of these questions to your team each week and write down the answers. That's your raw material, and it's what turns an AI draft into something only your MSP could publish.

Input Two: A Written Standard

The second input is a standard: a short written guide to how your company sounds.

Most MSPs have a brand voice only in the sense that the owner knows it when they hear it. That works when the owner writes everything. It stops working the moment a tech posts on LinkedIn or an AI tool drafts the newsletter.

A useful standard is specific enough to follow. It covers what you say about your work and what you'd never claim. It names the proof only you can offer, and it lists the words that are off limits. If "peace of mind" makes you wince, write that down.

Then give it to the tool. Most AI tools let you save context that carries into every draft, whether that's a Project in Claude or a custom GPT in ChatGPT. Load in your standard and a few examples of writing you're proud of. Every draft starts from your voice instead of the internet's.

Keep the Last Call Human

Even with good inputs, someone needs to read every piece before it goes out. Not to fix commas, but to ask one question.

The Test

Cover your logo and read the draft. Could a competitor have posted this?

If the answer is yes, it isn't ready. Go back to your raw material and find the detail only you could include, like the client situation or the phrase your ops lead actually uses. One real detail does more than three paragraphs of polished generalities.

This takes a few minutes per piece. It's also the step most MSPs skip, because the draft looks finished. It looks finished because AI is very good at making things look finished.

What AI Can and Can't Do for You

AI can handle the labor. It can turn a voice memo into a draft, or reshape one article into a month of LinkedIn posts on a week when nobody has time.

What it can't do is know your business. That part has to come from you.

So go back to that blog post about why small businesses need managed IT. The difference between that post and one only your MSP could publish isn't a better tool or a cleverer prompt. It's a ten-minute conversation with your ops lead and one page that says how you sound.

Give AI those two things and it stops pulling you toward the average. It starts helping you sound like yourself, more often than you could manage alone.

When's the last time you read one of your own posts and thought only your company could have written it?