What AI is actually good for in PR and marketing (and what it isn’t)

There are two unhelpful conversations about AI in marketing. One says it will replace the whole job. The other pretends it isn’t happening. Neither is much use if you run a brand and want to know what to do on Monday morning.

So here is what we have found, using it every day.

Where it helps

AI is very good at the parts of the job nobody went into PR to do. Turning interview notes into a clean transcript. Summarising a 40-page report so you know which three pages matter. Producing a rough first structure for a newsletter. Reformatting the same release for a trade title, a consumer title and a website. Checking copy against a house style. Pulling coverage into a report.

That is real time saved, and time saved on admin is time spent on the work that wins coverage: the idea, the angle, the call to the right journalist. Used like that, it makes a small team faster without making it worse.

Where it falls down

The problems start when AI is asked to do the thinking.

It has no point of view. It writes what is most likely, which means it writes what everyone else is writing. Ask for a brand voice and you will get one that would fit your nearest competitor just as well. Ask for expert comment and you will get something that sounds like comment and says nothing a journalist could use.

It gets facts wrong, confidently. A made-up statistic or a misremembered product detail in a press release doesn’t just get corrected. It costs you the journalist’s trust, and in a small industry, that travels.

And it is easy to spot. Editors are now receiving pitches and comment written by AI every day. They recognise the rhythm: the same phrases, the tidy lists of three, the lack of anything specific enough to be true. Having sat on the other side of that inbox for a long time, we can tell you what happens to it. It gets deleted.

What only people can do

Judgment. Deciding which story is worth telling, and which one will look good in a client meeting but never run. Knowing when to hold a launch for a better slot, and when a feature is a fit for one title and wrong for another.

Taste. Choosing the image that makes an editor stop scrolling. Knowing when a sentence is right, not just correct.

Relationships. A journalist picks up the phone because they know you, and because the last thing you sent them was worth their time. No tool builds that. It is earned over years, one useful email at a time.

And accountability. When something goes out under your name, someone needs to have read it, understood it and decided it is right. That should always be a person.

How we use it

Plainly: AI helps us with the mechanical parts of the job, and a person decides what goes out. Every release, every post, every line of brand copy is read and signed off by someone who knows your business and has a view on it. We don’t hide that we use the tools, and we don’t let them choose what we say.

If you are working out where AI fits in your own marketing, that’s the test we would suggest. Use it to give your people more time for the work that needs a person. If it is replacing the thinking instead of clearing space for it, you will see it in the results, and so will everyone reading.

Our full position is on our Where we stand on AI page.