Everyone Uses AI Now. Here’s How We Keep Our Content Human.

Most people reading a blog post these days assume a machine helped write it. Often they’re right. We’d rather say so than pretend otherwise, especially because we use AI ourselves. We use it, and we still expect every post to sound like a person wrote it, because one did.

What buyers do about it

TrustRadius’s 2026 research, which surveyed technology buyers, found that 63% used AI during their purchase journey. Of those, 94% said they fact-check its answers at least some of the time. A recap of a Forrester B2B Summit panel summed up the pattern simply: buyers ask AI first, then go looking for a human to confirm.

So the AI answer is the starting line and a person is the tiebreaker. If you publish things for a living, that’s a little uncomfortable. It should also change how you write. If a buyer is going to double-check you anyway, the content that wins is the content that holds up when they do.

Where we stand

AI isn’t the villain here. It’s a useful tool. The trouble is how much of what comes out sounds identical: smooth, correct, interchangeable. Read three vendors’ posts on the same topic and you often can’t tell which is which.

Quick test. Read your own post cold and ask whether a competitor could publish it without changing a word. If yes, it isn’t doing much for you.

We’ve seen this from the inside. A draft can come back clean and confident and still say nothing, and it takes a person to notice. These tools are good at sounding sure. They’re not good at knowing when they shouldn’t be.

How to spot autopilot writing

You don’t need a detector. The giveaways are easy to see:

• It opens with a sweeping line about how fast the world is changing
• Every paragraph is the same length and ends with a tidy little lesson
• Nothing in it could only have come from this company
• The examples are hypothetical, never specific
• Swap a few nouns and it could be about any industry

That’s how a reader sizes it up, and it’s the list we hold our own drafts against. Feel free to run it on this post. We’d rather you check than take our word for it.

How we use it

AI helps us move faster. We use it for research, first drafts, and getting past a blank page. Then people take over. Someone decides what the point is, checks the facts, rewrites whatever sounds off, and makes the final call before anything goes live. A claim we can’t back up comes out. A paragraph that sounds like nobody in particular wrote it gets rewritten until somebody did.

That last step matters more than it sounds. The edit is where a draft stops being generic and starts sounding like the people whose names are on it. It also means some posts take longer than they would if we published the first draft. We’re fine with that. Buyers can’t see how a post was made. All they have is how it reads, so how it reads is the whole job.

Some things we don’t hand to a model at all. Opinions, for one, since a model will confidently produce a take on anything, which is the problem. Stories from our own experience, because all a model can do is imitate them. And anything where being wrong costs something: a statistic, a legal point, a claim about what a client got. A person checks those against a source first.

What data misses

Data tells you what happened. It’s much weaker on why, and weaker still on what’s coming. Years in B2B teach you things that never reach a spreadsheet: which numbers to doubt, which prospects deserve patience, which “no” means “not yet.”

Anyone who’s spent time on the conference circuit has watched a prospect go from “not interested” to biggest advocate over a couple of years of hallway conversations. A lead score would have given up early. We tell one version of that story in our trade show post.

Or take a prospect who goes silent for six months. A dashboard marks them cold. Someone who has worked the industry knows budget cycles, knows the person’s boss might have changed, and knows to try again when the new fiscal year starts. Same data, different read.

That’s not an argument against data. We use plenty of it. It’s an argument for not letting a dashboard make the call on its own.

Our team has years of that, and it goes into our work. A model can summarize what’s been written about B2B marketing. It hasn’t sat through the follow-up calls. It’s never been ghosted by a client, for one thing.

What you can count on

What makes content trustworthy? A named person with a point of view who’ll defend it. Specifics only someone in the work would know. Original data from our own work, like the exhibitor lists we build. Experience a model can’t have. Claims we can back up, and no numbers we can’t source. None of it is complicated. It’s just slower than hitting generate, which is why so few people bother.

We’re telling you all this because it’s true, because buyers already assume the worst, and because it holds us to it. If we say a person checks everything, we’d better be doing it.

If something we publish reads like nobody wrote it, tell us. We’d rather hear it from you than from a bounce rate.