We're going to be honest about something upfront: this wasn't a stress-free, everything-worked-first-time sixty minutes. There were two prompts that missed, one Midjourney generation we discarded entirely, and a section of copy that needed rewriting by hand. But from the moment we saw the brief to the moment the page was live, sixty minutes is what it took. That's the number that matters.
The client was a local plumbing company — a sole trader who handles emergency call-outs and regular maintenance across a mid-sized city. They had no website, were relying entirely on word-of-mouth and a Google Business listing, and had a genuine business need: a dedicated landing page to capture leads from paid search. They needed it quickly, the budget was limited, and the brief was simple: make it clear what the business does, make it easy to call, make it work on mobile.
Here's exactly what we did.
The Brief
The brief arrived as a voice note and a few follow-up messages. Key points: "emergency plumber, available 24/7, covers [city area], no call-out fee, fast response, want people to call or fill in a form." That was essentially it.
Before touching any tool, we spent five minutes extracting the core message. The differentiators were the 24/7 availability, the no call-out fee (which is genuinely unusual in the space), and the fast response time. Everything else was table stakes. Those three things became the anchor points for the copy hierarchy.
Step 1 — Copy with Claude (12 minutes)
We went to Claude first. The prompt we used was direct:
Prompt: "Write a landing page for an emergency plumber. Key differentiators: 24/7 availability, no call-out fee, fast response. Target: homeowners and renters in a mid-sized UK city who have an urgent plumbing problem right now. Tone: calm, competent, trustworthy — not salesy. Include: headline, subheadline, 3 benefit statements with short copy, a social proof section placeholder, and a CTA. Keep it concise — every word should earn its place."
The first output was good but slightly generic. The headline was "Expert Plumbers, Available 24/7" — accurate, but not differentiated. We pushed back in one follow-up: "The headline doesn't lead with the most compelling differentiator, which is the no call-out fee. Revise the headline and subheadline to lead with that, and make the tone more like a reassuring expert speaking to someone who's stressed."
The revised headline: "No call-out fee. Here within the hour." Subheadline: "Emergency plumbing across [City] — 24 hours a day, 7 days a week. Fixed prices, no surprises." That was usable. The three benefit blocks were clean. The CTA — "Call now: [number]" and a secondary "Get a quote" — was exactly right for the conversion intent.
Total time including refinement: twelve minutes.
Step 2 — Structure with Relume (8 minutes)
Relume's sitemap and wireframe tool is underrated for single-page projects. For a landing page, you don't need a full sitemap — but the component library is excellent for quickly laying out section order without touching a design tool.
We used Relume to select the section sequence: hero with CTA, three-column benefit section, a trust bar with simple icons (24/7, no call-out fee, fast response as visual badges), a brief about section, a contact form section, and a footer. Eight minutes to get the wireframe skeleton right. This step is often skipped when people "just want to get building" — and it's almost always a mistake. Having the section order locked before you touch Framer saves fifteen minutes of rearranging.
Step 3 — Visuals with Midjourney (10 minutes)
The hero section needed an image. For a plumber, we wanted something that communicated competence and professionalism without being a clichéd wrench-in-hand shot. Our prompt:
Prompt: "Professional plumber working in a modern kitchen, calm and focused, close-up on hands and tools, clean natural light, documentary photography style, muted warm tones, shallow depth of field --ar 16:9 --style raw --v 6"
The first generation gave us four options. Two were unusable (face visible and slightly AI-off, wrong tool type). One was mediocre. One was exactly right — hands on pipework under a sink, warm available light, authentic feel. We took that one, cropped it to the hero dimensions, and that was the visual work done. Ten minutes including the generation wait time.
We also generated a simple abstract background texture for the CTA section — neutral, slightly warm, no subject. That was one generation, first output acceptable, two minutes.
Step 4 — Build in Framer (22 minutes)
Framer AI is genuinely impressive for getting a basic layout live quickly, but it has a ceiling. We used it to generate an initial frame from a text description of the page — "professional services landing page for emergency plumber, clean white background, dark navy accents, clear CTA hierarchy, mobile-first" — and got a frame that was 70% of the way there.
The twenty-two minutes broke down roughly as follows: five minutes getting the AI-generated frame into a state worth iterating from, twelve minutes replacing placeholder text with the Claude copy and dropping in the Midjourney images, and five minutes fixing responsiveness on mobile (the hero text stacked awkwardly on small screens and needed manual adjustment).
Framer AI gives you a very good starting point. Turning a good starting point into a finished page is still design work — it's just faster than starting from blank.
We did not use custom animations, complex interactions, or anything beyond the default hover states. Those would have added time and weren't relevant to the conversion goal of this page.
Step 5 — Publish (8 minutes)
Publishing in Framer is genuinely frictionless. We connected the client's domain (they already had one registered), configured the basic SEO metadata — page title, meta description, Open Graph image — and hit publish. The live URL was working within a few minutes of the domain connection propagating.
We set up a basic form connected to a Notion database so the client could see leads without needing a CRM. That was the only integration, and it took three minutes to configure.
What Took the Most Time (And What AI Still Can't Do Fast)
The honest accounting of where time went versus where we expected it to go:
- Mobile responsiveness was the biggest manual task — Framer's AI doesn't reliably produce good mobile layouts for every section
- The contact form styling needed significant manual adjustment; Framer's default form components are functional but not polished
- SEO metadata — writing a good meta description and page title still requires human judgement and wasn't something we delegated to AI
- The trust indicators (icons for the benefit bar) required sourcing from a free icon library; AI didn't generate these, we used Phosphor Icons
- Reviewing the final page on a real mobile device identified three things that looked fine in Framer's preview but didn't work on an actual iPhone — each fix was thirty seconds but they needed to be caught
The parts that would not have been feasible without AI: generating strong, on-brief copy in under fifteen minutes, having hero imagery that felt authentic and professional without a photography budget, and having a buildable component structure to work from rather than starting from blank.
The parts that were still clearly human work: strategic decisions about what to lead with, quality control at every stage, the judgement calls about what to cut, and the final review pass. AI accelerated execution; it didn't replace thinking.
For a lead generation page targeting search intent, that's exactly the right division of labour.
If you're curious about using no-code tools to build quickly, our piece on no-code and AI builds goes deeper on the tools. And for the broader strategy behind fast builds, read our thoughts on what clients actually want in 2025.