I wasn't an early adopter of AI. I was skeptical. Didn't think it could really do what I do—namely write. Then I found that I could generate a white paper draft in an afternoon and spend the next day tidying it up and verifying claims. This is when I started paying closer attention to what commentators were saying about AI capabilities in the development world—the coders, the practitioners, the technical people.
I'm not technical, but a few themes kept emerging that I understood perfectly well as a writer for tech firms: AI works best with clear, structured instructions. Context matters. The more precisely you define what you want—and for whom—the better the output.
This is specification writing
For much of the 1990s and into the early 2000s, many enterprise software organizations were shaped by plan-driven disciplines: define the requirements, model the system, document the interfaces, and specify important behaviors before implementation. Object-oriented analysis and design reinforced this mindset, encouraging teams to think in terms of objects, relationships, responsibilities, and contracts. The assumption was that clearer design would lead to better software.
Then Agile pushed back. Iterate fast, ship early. Strive for the minimum viable product. Treat the spec as a liability. For software development, Agile largely won.
But in many ways, AI is turning the page once again. The practitioners getting the best results are the ones writing the clearest specs—not code, but language. Define your inputs. Constrain your outputs. Specify your context. Think carefully about what you're asking before you ask it.
The brief, product truth, and content as infrastructure
For AI developers, spec writing is increasingly the art of directing AI systems well—giving them clear goals, relevant context, operating constraints, and criteria for success. Enterprise content teams have been writing something similar for years, but we've always used a different term: the creative brief.
What does a creative brief contain? A defined audience, explicit tone and voice, structured output format, and constraints on what to include and exclude. If you've written briefs, managed a style guide, or built a messaging framework for a product launch, you've been doing a version of specification writing for years.
This structured, consistent, disciplined approach to content generation is quietly turning content teams into the keepers of product truth. In the generative search era, when a buyer asks an AI system about your product category, what surfaces isn't your homepage. It's the structured, coherent, authoritative content your marketing team has produced and maintained.
If you've heard the term "content as infrastructure" in AI writing circles, this is what it means. Content is the underlying layer that the business depends on to establish product truth. It's what AI systems reference, what search surfaces, what sales teams pull from, what buyers use to understand what a product does. It's persistent, structural, and load-bearing—more like a data layer than a publishing operation.
The discipline gap
Not every content team has invested in editorial infrastructure the same way. Some have built real messaging frameworks and documented the reasoning behind them. Others run more on tribal knowledge—the institutional memory of one or two senior writers who just know the voice, the product, and what the company would or wouldn't say.
But what lives in someone's head isn't available to an AI system. And in the generative search era, this gap has consequences beyond internal workflows. Vague or undocumented brand voice results in inconsistent content—which leads to inconsistent answers when a potential buyer asks ChatGPT "what does this product do?"
This is properly seen as a brand problem. Consistency across everything a content team produces is more than good practice. Increasingly it's the foundation for how AI represents the brand.
Domain expertise in a new context
There's a pattern emerging in the AI development world. The people getting the best results know their domain deeply enough to direct AI precisely, catch what it gets wrong, and recognize when the output is merely plausible rather than good.
Enterprise content writers know what they need from AI. They know when the output is accurate, when it's on-brand, when it's structured for the right audience—and when it's none of those things. This is the judgment that comes from years of writing to briefs, managing brand voice, and producing content that has to hold up under scrutiny.
The writers who pair this judgment with genuine AI fluency are strongly positioned—because accumulated expertise is exactly what this moment rewards. The instinct, domain knowledge, and structured discipline are already there. What's new is the opportunity to put all three to work at a scale that wasn't possible before.
← Back to Blog