Corporate waffle had a perfectly good career before AI arrived. “Delighted to announce” was doing heavy lifting long before anyone opened a chatbot.
I’m pro-AI. I also think we’re letting ourselves off remarkably lightly when we blame the technology for writing that has nothing to say.
Take a company announcement about a new service. The brief asks for something professional, engaging and on-brand. It includes a product description and a list of features. Nobody explains what customers currently struggle with, why this approach is better, or which claim the business can actually defend.
What, exactly, are we expecting the AI to uncover?
It can produce fluent copy from a weak brief. That fluency can make the weakness harder to spot. The same is true when a capable human writer tidies up a muddled argument. I’d rather fix the argument than spend another afternoon replacing “innovative” with “forward-thinking”.
For a small business, this matters. Time spent polishing empty words is time you could spend finding the useful thing you have to say. AI can help with that work, provided you give it something solid to work with.
Here’s how I’d use it.
First, write the belief in a sentence you would be willing to say to a customer. Perhaps you believe businesses make buying unnecessarily complicated. Good. Explain where the friction is and what you’ve changed. “We’re passionate about customer experience” gives the reader considerably less to work with.
Then supply the approved evidence. Product details, current policies, attributed customer feedback you have permission to use. Keep the source and date alongside each fact. Be explicit about what is uncertain and what must stay out of the draft. A tone-of-voice guide cannot tell a model whether last year’s service promise still applies.
Next, make the AI challenge the material. Ask which sentences could belong to any competitor. Ask where a sceptical reader would demand proof. Ask what the customer needs to know before making a decision. I find this a much more interesting brief than “make it punchier”.
Finally, keep a named person responsible for the final words. They should check the claims against their sources and decide whether the company stands behind the argument. AI criticism can be useful; it can also be wrong. Giving it an editorial job doesn’t remove yours.
That approach leaves room for personality. A business can be calm, funny, blunt or detailed. It can hold a view that a competitor would disagree with. Consistency should make a company recognisable without forcing every message into the same shape.
That’s what “One company. One voice.” means to me: shared facts and recognisable judgement, adapted to the people reading.
I want AI to give smaller businesses the room to communicate with more confidence and substance. But we have to bring something to the arrangement.
Before asking whether a piece sounds like AI, I’d ask a more awkward question: would anybody notice if our competitor had published it?