crop_standard_chisel.jpg

A lesson from the year 2000

I wrote my first CMS in 2000 simply because the open source alternatives were too heavy, and anything commercially licensed would have cost an outrageous amount of moneyβ€”starting from €15k a year (around €40k today).

The project was a county website for a community of 35,000 people. It needed news, events, a poll, some file downloads, link catalogues, and a few basic pages. Rather than wrestling with ezPublish, Mambo, or TYPO3 and training someone to use them, it was faster to just build exactly what was needed. Since then, I have adapted more than a few open source platforms into working websites and built heavily on various systems, only to watch the same cycles repeat.

The endless cycle of magic words

For years, we have had one magic word. It was easy to sell, so sales teams pushed it for everything. We ended up forcing simple websites to wake up a server, run PHP, and query a database just to serve a page that had not changed in weeks. We treated a specific tool as a universal solution, which led to a decade of poorly planned integrations.

Today, AI is the new magic word. The sales pitch has simply shifted. Now the promise is that we will make it faster, we will make it AI, we will use AI for this and that. The current trend suggests we should abandon the traditional CMS entirely. You keep a site in GitHub, ask an AI to update the page, review the code, and publish. No databases, no PHP, just flat HTML and speed.

For a simple static site, this is brilliant. But because this shift is driven by sales and hype rather than engineering, we are in danger of blindly moving from bad legacy integrations straight into bad AI integrations. AI is just the newest shiny hammer, and using it blindly will create its own mess.

Professional assessment over the sales pitch

This is where true engineering separates itself from development driven by sales. Professionalism means assessing the actual task in front of you and forecasting possible developments, rather than giving in to the hype and going with the next new thing.

The original reason for a CMS was to give people without technical skills a reliable and structured way to edit a site. AI is not going to change that core requirement. If you give a non-technical marketing department a GitHub repository and an AI prompt, you might save on database hosting today. However, you will eventually pay a massive tax in workflow friction, broken layouts, and lost version control when they try to scale their operations.

A professional architect asks what happens when the editorial team grows. They ask how the system handles permissions, multilingual content, and dynamic integrations three years down the line. They anticipate the point where a collection of flat HTML files managed by AI prompts becomes a maintenance nightmare.

It is still about the content

If my experience since the turn of the millennium has taught me anything, it is that frameworks rise and fall, but the core mission remains untouched. We shifted from rendering on the server to headless architecture, to static generation, and now to coding assisted by AI.

But the foundation of the web remains exactly the same. It is still about the content.

The technology exists to serve that content to the user, not the other way around. A beautifully engineered stack is worthless if it makes publishing a chore, just as an easy publishing tool is worthless if it crashes under the weight of its own bloat.

In the end, it is professionalism that matters, not the stack. Stop chasing the magic words. Assess the actual needs of the project. Pick the right tools. Focus on the content.