Website Relaunch with AI

We had a website. Technically functional, factually correct — but it didn’t feel professional.

Built with WordPress and Elementor, developed in-house, grown over time. Yet a nagging feeling remained: the online presence didn’t match our standards.

As a consulting firm for Operational Excellence, we help organisations build structure, clarity, and professionalism. Our own website only partially conveyed that — the layout wasn’t consistently polished, the visual language wasn’t sharp enough, the overall impression solid but not distinctive.

That became the catalyst for everything that followed.

Why a Relaunch Was Necessary

WordPress is a proven system. Elementor promises design freedom. In practice, both also meant: learning the intricacies of themes and plugin dependencies, dealing with hosting and performance, managing constant update cycles.

Then there was the design aspect. Without formal training in web design, it quickly became clear: a functional website isn’t automatically a well-designed one. Inconsistent layouts, uncertainty about typography and spacing, colour decisions based on gut feeling — much of it was trial-and-error rather than systematic design.

Looking back, our first website was less a web project than a learning project. A significant share of the time went not into content but into technical details and design corrections.

The question was obvious: If AI can generate structures, content, and design proposals today — why not rethink the website from the ground up?

Don’t Optimise — Start Over

We could have reworked the existing site. New theme, better colours, tighter copy. But that wouldn’t have solved the fundamental constraints — neither the technical dependency on plugin ecosystems nor the lack of a systematic design foundation.

Instead, we chose an experiment: build a corporate website from scratch, driven entirely by AI. Not just generate a few texts with AI, but systematically develop the entire process — structure, navigation, messaging, visual direction, code — with AI at its core.

Our role changed fundamentally. Instead of manual layout tweaking and plugin fine-tuning, a new way of working emerged: brief, evaluate, refine. From executor to director.

13 Mockups in One Hour

The starting point was deliberately visual. Rather than drafting individual pages, we first had complete design mockups generated.

Within one hour, 13 distinct variants emerged — each with its own approach to layout, navigation, hero section, and typography. Some emphasised large-scale imagery, others typographic clarity. Some visually highlighted consulting expertise, others prioritised the service overview.

In a traditional design process, this exploration phase alone would have taken days or weeks. With AI, it became a focused selection exercise.

From the 13 variants, we chose a favourite — not purely on aesthetics but strategically: which mockup conveys our professional standards most clearly? With that decision, the design foundation was set. From this point on, we stopped “designing” and consistently built on the chosen mockup.

This solved one of the biggest problems of the previous approach: instead of fragmented one-off decisions, there was a coherent design logic from the very start.

From Mockup to Code

Implementation was done with Claude Code — an AI coding assistant that built the website not in a traditional CMS but as structured, generated code.

The approach was iterative. The same rhythm for every page:

Have AI propose the page structure
Develop content with AI support and sharpen the language
Adapt design elements along the chosen mockup
Evaluate, give feedback, refine — next page

Because the initial mockup had already set visual and content guardrails, later design decisions were significantly reduced. Instead of reinventing each page, we built on a defined framework and carried it through consistently.

The difference from the previous approach was tangible: less friction, faster decisions, more coherent results. What used to stretch across weeks — structure, copy, layout, polish — now came together in a fraction of the time.

The Tech Stack: Lean Instead of Heavy

The new website runs on a static architecture with Astro as the framework, hosted on Cloudflare Pages, and versioned via GitHub.

No CMS, no database, no plugin updates. That means a smaller attack surface and less maintenance overhead — but also a deliberate trade-off: every content change requires a code update and a fresh deployment. There’s no backend to log into and add content with a click. Publishing a new insight concretely means: create the file, commit, wait for deployment.

In return, previous hosting costs, the Elementor licence, and ongoing plugin maintenance all disappear. Monthly fixed costs dropped by around CHF 30. Not a lot — but combined with the more professional appearance and the significantly reduced security attack surface, it’s a clear improvement.

The setup is complemented by Cloudflare Workers for serverless functions such as newsletter sign-ups and Google Analytics for tracking user behaviour.

What We Learned

Looking back, this project was less of a relaunch and more of a shift in perspective — in how we approach digital solutions.

AI doesn’t replace strategic clarity. If you don’t know what your website should communicate, AI won’t give you a better answer. But if you have a clear vision, AI helps you realise it faster and more consistently.

The biggest time savings were in design. 13 mockups in one hour — that changes how you think about visual design. The question shifts from “How do I build this?” to “Which variant fits best?” Less construction, more selection.

The trade-off with a static setup is real. No CMS means more control, but also more technical involvement with every change. That’s a deliberate choice — but it’s not right for every team.

The role changes fundamentally. You stop implementing every detail yourself and start steering and evaluating outcomes. That requires a different skill set than traditional web development — less craft, more judgement.

The key takeaway: AI is not a substitute for professionalism. But it’s a powerful lever to make professionalism visible more efficiently — if you’re willing to fundamentally change how you work.