We will never build another WordPress website

I think we are done with WordPress. I have been building WordPress websites for 17 years. Roughly half of our clients still run on it today, and we will keep supporting them for as long as they want to stay. But I do not believe we will ever build another WordPress website again. That is a strange sentence to write after 17 years, so I want to walk through why we are done, what we are switching to, and how you should be thinking about the platform your own website is built on.

WordPress is not AI native, and no plugin changes that

The main issue with WordPress is simple: it is not AI native. Yes, I know you can connect AI to WordPress. I know there are MCP servers for it. You still cannot tell me it is AI native. Claude cannot just build you a WordPress website the way it can build you an HTML website. The theme layer, the plugin stack, and the database all sit between the model and the page. When you connect AI to WordPress, it fights you every step of the way. Anyone who has actually tried it knows exactly what I mean.

We spent six months auditioning replacements

So over the last six months we looked hard at the AI native CMS options. I was hopeful about EmDash, the open source WordPress successor Cloudflare released this spring. We tried Payload. Then I went and built my own CMS. It was obvious that WordPress had entered its final act. We just did not know what we would be switching to.

Fable was the unlock

The real unlock for us was Fable. Anthropic's new model is so good at software engineering that it changes the math on what a small team can maintain. We have built a ton of software with it, and one of those things turned out to be our dream CMS. I spun up our ideal system in a weekend. It has every feature we cared about from WordPress, including a drag and drop page builder, but the pages it produces are plain HTML on Next.js. It plugs directly into the knowledge base we keep for each client on GitHub. I can migrate any existing website into it without much drama. Everything lives on Vercel, so our clients still own their websites. We just get access. And because the site itself is nothing but HTML and JavaScript files, moving to another CMS someday would be easy. I honestly do not think we will need to.

Maintaining your own CMS used to be crazy

Five years ago, an agency writing its own CMS was itself a red flag. It meant lock-in for the client and a maintenance bill that never ended. AI has flipped that. Software is now cheap enough to build, and cheap enough to keep running, that owning the exact tool you need makes more sense than renting an approximation of it. We have a small team of AI native developers looking after ours, and over the next six months I imagine we will recommend that many of our clients make the switch.

One caveat. If you run an e-commerce business, I would wait. WooCommerce is creaky, but payments, inventory, and checkout are exactly where you do not want to be anyone's early adopter. It will be a little longer before we tell stores to jump.

How to think about your own platform

If you are deciding where your website should live, the question is no longer which CMS has the biggest plugin directory. The question is whether an AI can work on your site without fighting the platform. Ask what the output is. If the answer is plain HTML and JavaScript files you can pick up and move, you are safe almost anywhere. And ask who owns the hosting and the deployment, because you should own it, not your agency. That last rule has been true for 17 years. The other two are new, and they are the reason that, for us, WordPress is finally over.

Opinion columns reflect the personal views of the author. Our reporting on the stories referenced here lives on the linked pages.