If you’ve been following my recent emails, you’ll know I’ve spent the last several weeks building and refining a fairly substantial web application.
One of the first things I wanted to get right was its content management system.
I knew there would be ongoing editorial work, data updates, corrections and all the other little jobs that come with running a website. I didn’t want to spend my days wrestling with an interface just to make relatively simple changes.
So I built my own CMS.
Not because I wanted to recreate WordPress or another existing platform. Quite the opposite.
I wanted something built around the way I actually work, with the controls I need and very little else.
But there was another reason for keeping it simple.
I wanted AI agents to be able to use it with minimal additional work.
Building for people, with AI in mind
From the outset, I wanted the CMS to be usable by me, VAs, and eventually by AI agents operating under supervision.
That influenced how I approached the interface.
Straightforward navigation. Predictable forms. Clear editing workflows. Controls that do what they say.
Nothing particularly revolutionary.
And it fits my broader approach to development.
I’ve become increasingly sceptical of the tendency to reach for complicated frameworks, abstraction layers and sprawling technology stacks before establishing whether they’re necessary.