The most common complaint about AI content is that it's generic. And it is — when you give it a generic brief. 'Write a 1,000-word article about content marketing' will produce a 1,000-word article about content marketing that could have been written by any AI, about any company, for any audience. That's not the tool's failure. It's the brief's failure.
Prompting AI for content is a skill that most teams haven't invested in learning. They try it once, get a mediocre output, decide AI isn't ready, and go back to writing everything from scratch. Or they accept the generic output, dress it up a bit, and publish it — which is where the 'AI content is bad' reputation actually comes from. Neither of those is right. What's missing is the brief.
Start With the Reader, Not the Topic
Before you type a word of your brief, answer these two questions: Who is reading this article, specifically? And what do they believe right now that this article should change or confirm? Not 'content marketers' — that's not specific enough. 'In-house content managers at B2B SaaS companies who've tried AI tools and been disappointed with the output.' That level of specificity changes the entire draft.
The belief question is even more important. Most articles are written to inform. The better ones are written to shift a specific belief or confirm a suspicion the reader already has. 'People who read this already know AI content can be bad. I want them to leave believing the problem is the brief, not the tool.' That's a thesis. A thesis produces a draft that argues something. A topic produces a draft that explains something. Arguments are more interesting to read.
The Elements of a Brief That Actually Works
Angle: Not just the topic, but the specific take. 'AI content is generic' is a topic. 'Generic AI content is a brief problem, not a tool problem' is an angle. The angle is the sentence that tells the AI what argument to make. Without it, the AI will hedge, balance, and produce a 'comprehensive overview' that doesn't commit to anything.
Audience: Specific enough that the AI can make vocabulary and assumption choices. A brief that says 'intermediate SEO practitioners who already understand keyword research' will produce different language choices than one that says 'small business owners new to SEO.' AI models adapt their register and assumed knowledge based on audience signals — use that.
Evidence requirements: Tell it what kinds of claims need sources. 'Any statistics should be from 2023 or later, cited with publication name' will produce a draft that flags or includes real citations. builds this in automatically — the source layer is part of the generation, not an afterthought. When citations are an afterthought, they're either missing or hallucinated.
Tone reference: 'Write like The Atlantic, not like a university textbook' is more useful than 'professional but approachable.' AI can sample from a vast range of tones. Give it a reference point it can work from rather than a vague adjective that means different things to different people.
What to Include That Most Briefs Skip
Common mistakes to avoid: tell the AI what the obvious, boring version of this article looks like — the one that leads with a definition and ends with 'in conclusion.' If you describe what you don't want, the AI will actively work around it. This is one of the most underused brief techniques.
Things the reader already knows: if your audience is experienced, telling the AI what to skip saves you from getting an article that over-explains basics. 'Assume the reader already knows what a content calendar is' removes the obligatory definition paragraph that would otherwise appear in the first 200 words. Every paragraph that explains something the reader already knows is a paragraph that loses them.
A specific structural requirement: not word count (that's the least useful constraint), but sections. 'Cover the brief elements, then the common mistakes, then a workflow they can use this week' is more useful than '1,200 words.' AI will fill a word count. It will also follow a structure — and a structure that matches your reader's mental model produces a much more useful draft.
Iteration Is Part of the Process
The first draft is a starting point, not a submission. Even a great brief produces a draft that needs human editing — the voice, the specific examples, the first-person experience signals that only a real person can add. Plan for a revision pass. What you should not be doing is writing the article from scratch because the AI gave you something generic. If that's happening consistently, the brief needs work, not the tool.
A practical test: hand your brief to a colleague and ask them to tell you what article they'd expect to get from it. If they describe something generic, your brief is generic. If they can describe a specific, opinionated piece that couldn't have been written about any other company or audience — you're close.
The writer who learns to brief well gets better AI outputs and also becomes a better editor. The two skills compound.