How to Write and Mock Up a New Service Page With AI (No Copywriter or Designer Needed)

I got the brief at 7:30pm on a Thursday.

At Brightly, we were testing new programmes at speed — the gap between a product being ready and it needing to be live on the website, so sales could actually talk about it, was usually a matter of weeks. This time it was a weekend. We had a piece coming out in the NZ Herald the following Wednesday, the journalist needed the final copy by Monday, and the whole point of the timing was for the PR moment and the live page to land together. One without the other is half the impact.

I didn't have a team. I didn't have an agency on retainer. And even if I did, that turnaround either comes at a huge premium or it just doesn't happen — nobody's briefing a copywriter and a designer separately at 7:30pm on a Thursday and getting a polished page back by Monday.

So I didn't. I built it myself, in one sitting, with AI doing the heavy lifting on both the words and the layout.

Why the normal way doesn't work under real time pressure

The standard process is two handoffs, not one: brief a copywriter, wait for a draft, brief a designer separately, wait for a layout, then reconcile the two when the copy doesn't quite fit the design or the design undersells the copy. Every round of feedback means going back to two different people. None of that back-and-forth is really about the thinking — it's coordination overhead, and coordination overhead is exactly what you don't have on a Thursday night with a Monday deadline.

The workflow

1. Give it everything, not a one-liner.

My first instinct, like most people's, was to keep the prompt short — describe the task simply, because surely a huge wall of context would overwhelm it. That was wrong, and it showed: generic copy, forgettable, nothing you'd want your name on. What actually works is the opposite. I upload the full internal documentation on the product — everything the team has — plus the brand style guide and content style guide, before asking for a first draft. AI chatbots handle complexity well now. The more informed the first attempt, the less "rough" it is, and the more time I spend refining something good rather than rescuing something generic.

2. Be explicit about the audience and the goal.

Not just "who's it for" in the abstract — the actual segment, and what this specific page needs to do for them. This part doesn't change project to project. It's the one thing I always include, no matter how much or how little source material I'm working with.

3. Hand over whatever brand reference you actually have.

Sometimes that's a robust product brief from the team, ready to go. Sometimes the product itself is still half a thought, and the marketing copy I'm drafting ends up feeding back into pricing or feature decisions on the product side. Either way, I paste in whatever's genuinely useful — there's no fixed set of "always these three pages." It depends on what exists.

4. Get the words right before you look at layout.

Review the copy on its own first. Don't let a nice layout convince you mediocre copy is fine, and don't let rough copy make a good layout feel unfinished. These may feel like it contradicts the next step, but it pays to look at what you are producing from both angles.

5. Ask for a rendered mockup — not just text.

This is the instruction that changes everything, and it's the one people skip. I trained as a designer before I moved into marketing, and copy that reads as "right" on a page of plain text can be completely wrong once you see it sitting in its actual visual context — wrong length for the hero, wrong emphasis for the layout, wrong rhythm against the rest of the page. Nothing should get signed off outside the context it'll actually be seen in. So I explicitly ask for a working HTML page, not a copy document.

6. Iterate with an editor's eye, and trust your judgement on "done."

There's no fixed number of rounds. It depends entirely on how strong your source material was and how well you know the audience you're writing for. The skill that matters most here isn't prompting — it's editing. Knowing what's actually wrong with a draft, and being precise about fixing it, is what makes the iteration fast instead of endless.

What it actually replaced

That page went live in time for the Wednesday piece. Sales could point to it the moment the article ran. No agency could have turned that around on that timeline at any reasonable cost — and even an in-house copywriter-plus-designer combo would have struggled to hit Monday, let alone leave room for revisions.

The part this workflow doesn't solve for you

I want to be straight about a limitation here, because the brand of marketing I care about is the honest kind, not the hype kind.

A rendered HTML mockup is brilliant for getting the copy and layout right, and it's more than enough to get stakeholder buy-in — that's the whole point. But a mockup is not a published page. Getting from "here's a working HTML prototype" to "this is live on our actual website" is a separate step, and it doesn't always slot neatly into a live CMS.

I can close that gap myself, because I've also worked as a website developer — I build directly in WordPress and HubSpot, so I can take a mockup like this and have it live myself, without handing it to anyone. That's not a typical marketer's skill set, and I don't want to pretend it is. If you don't have hands-on CMS experience, you'll likely still need a developer, or at least someone comfortable in your specific platform, to carry the final version across that line. AI can get you a genuinely great mockup in one sitting. It can't guarantee that mockup drops cleanly into whatever your CMS happens to be doing that day.

"But surely the copy and design end up worse than hiring pros"

I get why people assume that. Here's my honest answer: I'm a qualified designer — I finished my design degree in 2017 and worked as one for years before I moved into marketing — and I've also worked as a website developer, building in both WordPress and HubSpot. So I can tell you directly — sometimes what the AI produces on a first pass is genuinely bad. That's real.

But give it a strong brand style guide and established templates, and it's excellent at matching that style and rolling it out at scale, consistently, over and over. I wouldn't tell anyone to let AI replace expert judgement outright. What actually happens is the job changes shape: the expert work becomes building the brand and the system in the first place — the templates, the guide, the standards. The repetitive part, rolling that system out again and again for every new page, every new service, every new segment — that's the part that gets to disappear. The work left over is the interesting part. Innovation, not repetition.

This is one of seven workflows in The Website & Product Content Playbook, my free guide to building, auditing, and fixing website and product copy without a content team.

Get the full playbook →