The AI problem most brands have
AI writing tools can produce a product description in 10 seconds. The problem is that the description will sound like every other product description in your category. "Crafted with care from premium materials," "perfect for everyday use," "makes a wonderful gift" — the phrases AI defaults to are the phrases that mean nothing to anyone.
This isn't a flaw in the tools. It's a flaw in how they're used. AI writing without input from your brand voice is like hiring a ghostwriter and giving them no brief. They'll write something technically correct and entirely generic.
What AI is actually good at
Before talking about limitations, let's be clear about where AI genuinely helps:
- First drafts. Writing is hard when the cursor is blinking. AI eliminates the blank page.
- Volume. If you need 20 product descriptions, AI gets you to a working draft of all 20 faster than you can write 2.
- Variations. Need a shorter version of this caption? A more formal version of this email? AI handles this in seconds.
- Structure. AI is good at organising information, writing outlines, and suggesting angles you hadn't considered.
The input determines the output
The difference between generic AI output and output that sounds like your brand is the brief you give it.
Instead of: "Write an Instagram caption for my handmade ceramics business."
Try: "Write an Instagram caption for a handmade ceramics business based in Lisbon. Our voice is warm but understated — we don't use exclamation points or words like 'amazing.' Our customers are home-focused people who care about where things come from. The post is about a new mug collection in earthy greens. Keep it to 3 sentences."
That brief takes 30 extra seconds. The output gap is enormous.
Train it with your own examples
The most effective technique: paste in 3–5 examples of your best existing content and ask the AI to "match the voice and style" of those examples when writing something new.
You're essentially showing it what you actually sound like, rather than hoping it guesses correctly. This works remarkably well for captions, product descriptions, and email copy.
Your role is editor, not approver
Don't look at AI output as something to accept or reject wholesale. Look at it as a first draft that needs your edit. Take the structure, the idea, or the angle — then rewrite the specific phrases to sound like you.
Often the most useful thing AI does is confirm an approach you were already considering, or surface an angle you'd have found eventually but faster. You're not delegating your voice. You're using a faster path to the draft so you can spend your editing time on the parts that actually matter.
What AI can't do for your brand
AI cannot know your actual story. It doesn't know why you started, what went wrong last year, what you noticed at the market on Saturday, or what your best customer said to you last week. All of that is yours.
The brands that use AI well treat it as a capable but context-free assistant. They feed it specifics. They correct the generic phrases. They inject the details that only they know.
Used this way, AI is genuinely useful. Used without that discipline, it makes every brand sound the same — and brands that sound the same compete only on price.