AI set up around how your business works — the rules in order, the team on board, and a way of working you manage yourself afterwards.
It's already happening
Somewhere in your business, someone is typing into an AI window right now.
Everyone uses it by now. A chatbot on the site, text rolling out of some window, a colleague pasting customer data into a tool without a second thought. Since 2 August 2026 you have to be able to show how that's arranged at your place. But getting the rules in order is the boring part. The interesting part is what becomes possible afterwards: work that does itself, and systems you change with an instruction in plain language instead of with a mouse.
Since August 2026 the AI Act asks three things of virtually every business that uses AI: an overview of which tools are running, a policy on who may do what, and staff who demonstrably know the risks. That sounds heavier than it is. A register, a policy document and an hour-and-a-half session cover it for most businesses.
Which tool, which data, and where it runs. Those three questions determine whether AI moves you forward or gets you into trouble. Customer data doesn't belong in a window nobody knows the destination of. So the setup starts with what may not go out, and builds around that.
Demonstrable AI literacy has become a requirement, but that's not the reason to do it. Whoever understands what a model can and can't do gets more out of it and falls into fewer traps. Explanation about your own work, with your own examples — not a course with screenshots of some other company.
The end product isn't a tool but a way of working. A structure that's right, rules that are fixed, and a site or system you then change with an instruction in plain language. SEO, schema and load speed are baked in — no separate optimisation round afterwards. And little is left to attack: no database, no login screen, no third-party plugins. For websites that's a package with a name and a price: AI CMS, from € 6,750.
This site is the testing ground. Every page, every piece of data and every rule here is set up to be understood and changed by an AI — and that's exactly what happens. No CMS, no plugins, no update button. Plain PHP and JSON, with a structure and a set of rules the next person working on it understands right away. Human or machine.
That's the difference with an advisory report: what I propose is already running. And because SEO, schema and speed sit in the way of working instead of in a separate round afterwards, there's nothing to tinker with later.
If you use AI: yes. Since 2 August 2026 the AI Act applies to virtually every company that deploys AI, regardless of size. For a small business it comes down to three things: knowing which tools you use, a policy on who's allowed to do what, and staff who are demonstrably informed. The heavy requirements for high-risk applications have been postponed to December 2027.
No, that's exactly the point. The setup and the rules come with it, so your team can do it themselves. Questions are always welcome, but you're not tied to anything.
Then everything stays in place. What you get is plain PHP and JSON: readable without any tool, editable by any developer. AI makes the management easier, but nothing depends on it.
That depends on which tool, and where it runs. The principle is simple: data you don't want out in the open doesn't go into a service whose processing you don't know. During the setup you decide per type of data what's allowed where, with a data processing agreement where that's needed.
Better than you're used to. Most sites are hacked through an outdated plugin, a weak password on a login screen or a hole in the database — and those three aren't here. What's left is the server and the PHP version, and that's a manageable list that gets kept up to date.
Especially so. The smaller the team, the more work is left lying around that a good setup picks up. And the law makes no exception for size.
Since 2 August 2026 the AI Act applies to virtually every business. What must you be able to demonstrate you've arranged, and what can still wait?
AIWhat does implementing AI cost in 2026? From an afternoon of policy to a fully set-up way of working — what drives the price and where the money leaks away.
AIWordPress and Drupal are building AI in as fast as they can. But what happens when the AI no longer needs the dashboard? Why the CMS structure is being torn up — and what replaces it.
