I rebuilt the way I build websites from scratch

The moment when updating was no longer enough

For years, continuity in my work also came from relationships built over time. Returning clients, people who knew me, and collaborations that did not require me to explain who I was and how I worked from the beginning every time.

Those relationships have not disappeared. Some clients still trust me with their work, and that trust is worth more than any slogan. But that continuity also exposed a weakness: for a long time, I had not needed a system capable of reaching new people.

I found myself facing an uncomfortable question: was the way I presented, produced and sold my work still suited to the market I was trying to remain part of?

I could have simply redesigned my website, changed a few lines of copy and updated the portfolio. It would have been easier, but it would have preserved the same system that was no longer creating a future.

The problem was not an outdated homepage. I needed to rebuild how I designed, demonstrated, delivered and maintained a website.

The easier route would have been to chase the market

Publishing a web page is remarkably easy today. CMS platforms, themes, builders and AI tools can produce something that looks like a website within hours. For limited needs, that can be enough; pretending otherwise would be dishonest.

I could have responded by offering quickly assembled sites, adding services and plugins on demand, or competing on price with people who had just entered the field.

I did not, because that would not have been an evolution of my work. It would have meant giving up the part where thirty years of experience can still matter: understanding the problem, choosing the architecture, reducing dependencies, verifying the result and standing behind it personally.

Starting from scratch does not mean becoming a beginner again

In 1995, a page might begin in Photoshop, be sliced up and reconstructed with an HTML table. Since then I have worked with programming languages, databases, animation, CMS platforms, ecommerce, responsive design, external services and AI tools. I have seen supposedly definitive solutions become obsolete, while others dismissed as outdated remained proportionate to the problem.

That history does not make me right by virtue of seniority. It does help me recognise a recurring mistake: confusing a tool’s novelty with the quality of the project.

When I started again, I did not return to the beginning. I used what I had learned to build a new system without carrying every old habit with me.

What I actually rebuilt

The visible result is a new website. Most of the work, however, lives beneath and around the pages.

These tools were not built to add more items to my offer. They start with practical problems: understanding before spending, reducing the risk of a decision, and allowing the parts that change to be updated without turning every site into a general-purpose CMS.

The least visible part of the rebuild is precisely this: every solution must have a reason to exist, a clear scope and a way to verify the outcome.

AI has not replaced my work; it has expanded its reach

I use AI every day. It helps me compare large amounts of information, check consistency, prepare tests, analyse data, document decisions and find errors that would otherwise take far longer to uncover.

I do not hand over responsibility for the result. Plausible copy can be inaccurate. A working component can be inaccessible. An attractive page can carry the wrong message. A test passed locally can fail in the real environment.

For me, AI is not a return to mass-produced websites. It is the tool that allows me, working alone, to bring a level of analysis and control to each project that was previously out of reach.

What I do not intend to change

I cannot control which article, conversation or project will produce the next opportunity. I can control the quality of what I show, the clarity of the offer and how I respond when that opportunity arrives.

I will not sell a website as a black box. I will not register a client’s domain or services in my name. I will not present an automated score as a certification. I will not add a technology simply because it is fashionable. I will not frame AI as a shortcut capable of replacing goals, content and responsibility.

I can change tools, my offer and how I communicate. I will not change the obligation to deliver a product the client understands, owns and can verify in their specific case.

What changes for someone choosing a website

Someone who entrusts a project to me should not need to care about the number of files, the programming language I use or how my Git workflow operates. They should be able to get answers to simpler questions:

• who designs and who develops;
• what is actually included;
• who will own the domain, accounts, content and data;
• how performance, accessibility and security will be checked;
• what happens after launch;
• which tools and providers the website depends on;
• who takes responsibility when something does not work.

I rebuilt the way I work so I can answer these questions before asking for trust.

I am documenting this rebuild before knowing its outcome. That is less comfortable than retelling it after the result is secure, but it is also a more honest way to show how I work.