AI Is Your New CMS

A content management system exists to solve one problem: websites are made of code, and most of the people who need to change them don't write code. So we put a dashboard in front of the code. You log in, find the right page, work the editor, and hope the layout holds when you hit publish.

For twenty years that trade-off made complete sense, and WordPress made it better than anyone — which is why it runs so much of the web. But it's worth being clear about what you were ever actually buying. Nobody wakes up wanting a dashboard. You wanted the new pricing live before the promo started. The dashboard was always the means, never the point.

Which raises a question that didn't exist a couple of years ago: what happens when the code itself becomes something you can just talk to?

AI-assisted development tools can now read an entire website's codebase and hold all of it in working memory at once. That's the shift. It means you can hand the site to an AI the way you'd hand it to a developer: "update the team page, Sarah's joined as an account manager" — and it finds the right file, makes the edit in the site's existing style, and stages it for review.

This is how northbase.au runs day to day. The site is a custom build with the content living directly in the code. When we publish a case study, rework a service page, or fix a typo, the workflow is the same: describe the change, AI makes it, a person reviews it, and the site deploys automatically the moment the change is approved.

The first thing you notice is everything you stop doing. There are no plugins to update, no CMS security patches to stay on top of, no database to back up and maintain. A large share of what a website maintenance plan traditionally covers simply has no equivalent here, because the attack surface a CMS carries isn't there.

The second is speed. A site that's pure code, with no CMS assembling pages on the fly, stays as fast as the day it launched. Ours scores 99 for performance on desktop in Google's Lighthouse — with straight 100s for accessibility, best practices, and SEO — and that isn't a launch-week number that decays as plugins pile up. There's nothing to pile up.

The third is quieter but might matter most: every change is tracked in version control. Each edit is recorded — what changed, when, and why — and any change can be rolled back to the exact previous state in minutes. Most CMS setups push edits live the moment you hit update; this workflow has a review step and a full undo history built in.

Now the part most articles like this leave out. If your team publishes five posts a day, runs a dozen content contributors, or needs to make changes at 9pm on a Sunday without anyone else involved, you want a CMS — and a well-managed WordPress site is still the best version of that. We manage plenty of them, and we're not expecting that to change.

AI-managed content suits a different rhythm: sites that change weekly or monthly, through an ongoing relationship with a developer or agency. The costs move too — you're paying for that relationship rather than for licences and patching, which works out well at this cadence and wouldn't at a newsroom's.

And it's early days. The workflow needs a human review step, and in our view it should keep one. The AI does the labour; a person stays accountable for what goes live.

For years, choosing a website platform meant choosing a dashboard. The question we now ask clients is different: how often does your site really change, who makes those changes, and how fast do they need to be live? Answer those three honestly and the platform decision mostly makes itself.

Some of those answers still lead to WordPress. More of them than you'd expect now lead to a fast, code-first build with AI-assisted content management behind it — and when the site needs to do real work, to a full web application. Either way, you're no longer choosing a platform because it's the only way non-developers can change a website. That constraint is gone, and it was the constraint holding the whole decision in place.