AI writing sounds generic when it's told nothing specific. What to write down about your brand so it sounds like you, and the one rule to keep absolute.

A row of blue and pink book spines on a shelf, one pale yellow book pulled forward and glowing

If you've asked an AI to write a LinkedIn post for your business, you've probably had the same reaction most people do.

It's fine. It's grammatical. And it could have been written for anyone.

The usual conclusion is that AI writing is generic. The better explanation is that the AI wasn't told anything specific, so it filled the gaps with the average of everything it has read.

The missing piece is context

An AI model knows a lot about writing in general and nothing about you. It doesn't know who your clients are, what you believe that your competitors don't, which words make you wince, or which claims you can actually back up.

Give it none of that and it falls back on the safest possible version. Give it the right context and the same tool starts to sound like a business with a point of view.

That gap, between what the tool can do and what it's been told, is where most of the quality is decided.

Your brand, written down, is now a working asset

Most teams now produce more content than they have writers for: posts, pages, emails, proposals. AI makes that possible, but only if it has something solid to work from.

So the written description of a brand has become a working tool, used every day. It shapes every draft, whoever or whatever writes it. A clear one keeps a small team consistent across every channel. A missing one means every request starts from zero and comes back sounding like everyone else.

Strict where it matters, open everywhere else

It's tempting to respond by writing rules for everything: every phrase to avoid, every sentence shape, every word swap. That usually makes things worse. An AI working through a long checklist writes to the checklist. The results come out stiff, and after a few weeks every post sounds the same.

The better balance is a few firm rules where a mistake would cost you trust, and room everywhere else. Spelling conventions, words you never use and facts you can stand behind deserve hard rules. Voice works better described than dictated: what you want readers to feel, and a few pieces of writing you're proud of. Then let the AI find the sentences.

The one rule that should be absolute

If there's one line that belongs in every brand document, it's this: never invent a client story, a quote, a statistic or a result.

An AI with no facts will produce plausible ones. A testimonial that sounds real. A percentage that feels right. On a website, that's the fastest way to lose trust, because the one reader who knows it isn't true is usually the one you most wanted to impress.

So a good brand document has two halves: how you sound, and what's true. The second half is a list of real facts the AI may use, and a clear instruction to leave a gap rather than fill it.

What goes in one

If you want to write your own, start with:

  • Who you're for. Your actual clients, described the way you'd describe them to a friend.
  • What you believe. Three to five positions you hold that not everyone in your field does.
  • How you sound. A short description and a few examples of writing that feels like you. Examples, not a rulebook.
  • What you never say. A short list of words, phrases and claims. Keep it short.
  • What's true. Real results, clients you're allowed to name, real quotes. Everything else stays out.

The takeaway

Before asking an AI for one more post, spend an afternoon writing down who you're for, how you sound and what's true. Keep the rules few and firm, and leave the rest open.

Your AI output is only as good as your brand context. And unlike the tools, which change every few months, that document keeps working whichever one you use.