In advertising, the brief is everything. It is the single most important document in the entire process. A good brief doesn't tell the creative team what to make. It tells them what problem to solve, who they're solving it for, what the world should feel like when they're done, and what success looks like. It sets constraints. It provides clarity. It gives direction without dictating the answer.
A prompt is a brief. I realised this about three weeks into using Claude Code, and it changed everything.
I'd spent fifteen years writing creative briefs at agencies like Poke, Dare, AnalogFolk, and Monks. Distilling complex business problems into single pages that unlocked great work. Learning the hard way that a vague brief produces vague output. That an overly prescriptive brief kills creativity. That the best briefs sit in a precise sweet spot: clear enough to guide, open enough to inspire.
That skill translates directly to working with AI. Not loosely. Not metaphorically. Directly.
The brief-writing muscle
When most people first use an AI coding tool, they do one of two things. They either give it almost nothing -- "build me an app" -- and get back something generic and useless. Or they try to micromanage every line, describing exactly what code to write, which defeats the purpose entirely because if you knew the code, you wouldn't need the tool.
The sweet spot is the brief. You describe the problem, the audience, the feeling, the constraints. You say what it should do, not how to do it. You provide context about why it matters. You set guardrails without dictating the execution. This is literally what strategy directors do for a living.
When I'm building a product with Claude Code, the prompt I write reads almost identically to a creative brief I might have written for a design team. Here's the audience. Here's the problem they have. Here's what good looks like. Here are things to avoid. Here's the tone. Here's what success means. The only difference is the recipient -- an AI instead of a copywriter or art director.
Clarity is the skill
The thing people get wrong about AI is thinking the skill is technical. It isn't. The skill is articulation. Can you describe what you want clearly enough that something else can build it? Can you identify the constraints that matter and ignore the ones that don't? Can you provide enough direction to be useful without so much that you strangle the output?
This is not a coding skill. This is a communication skill. And strategy directors have been practising it for decades.
I can't write a line of JavaScript from memory. I wouldn't know a React component from a Vue directive. But I can describe, with extreme precision, what a product should do, who it's for, how it should feel, and what constraints matter. That turns out to be enough. More than enough, actually. It turns out to be the hard part.
Taste as direction
Here's where it gets interesting. A brief isn't just clarity about the problem. A great brief also embeds taste. It communicates not just what needs to happen, but what quality looks like. It has a point of view about what good means.
When I tell Claude Code to build something, I'm not just describing functionality. I'm describing an aesthetic. "Clean, not busy. Confident, not loud. Typographically led. Think Monocle's editorial layout meets Bloomberg's data design." That kind of direction changes the output completely. The same functional requirements, with different taste direction, produce wildly different products.
This is exactly how it works in agencies. The same brief, given to two different creative teams, produces two completely different campaigns. The brief provides the problem and constraints. The taste of the person directing it determines the quality. An AI tool works the same way. It can build anything -- but what it builds depends entirely on the taste of the person directing it.
The career wasn't a detour
People sometimes ask if I regret not learning to code earlier. If all those years in advertising were time I could have spent becoming a developer. Absolutely not. Those years were the training.
I learned to distil complex problems into simple briefs. I learned to direct creative talent without doing the work myself. I learned what good looks like across dozens of categories, brands, and media. I learned to iterate, to give feedback that makes work better without destroying it, to know when something is almost right versus fundamentally wrong.
Every single one of those skills applies to working with AI. The strategy director who can write a brilliant creative brief can write a brilliant prompt. The one who can give precise, constructive feedback to a designer can give the same feedback to Claude Code. The one who knows what "good" feels like across different contexts can direct AI output toward quality that generic prompting never achieves.
The people who will be best at directing AI aren't the ones who understand how it works technically. They're the ones who understand how to communicate what they want. Who have spent years practising the art of translating vision into instructions. Who know that the gap between a mediocre brief and a brilliant one isn't information -- it's clarity, taste, and point of view.
Strategy directors make the best AI directors. Not because we understand the technology. But because we've spent our entire careers doing exactly this: telling talented entities what to build, and knowing what good looks like when it comes back.
The prompt is the brief. And I've been writing briefs for fifteen years.
