Blog Blog · July 15, 2026 · 2 min read

How we used AI to refactor our WordPress taxonomies

AI for WordPress

Our blog has fifteen years of history in it, and until last week the categories showed it. Twenty categories, fifty-five tags, going back to 2007. Magento, Joomla, osCommerce, a catch-all bucket called Stream. Good history, but it had drifted a long way from the work we do now.

So we rebuilt it. In one working session I sat with an agent, Claude Code wired to our WordPress over the REST API, and we refactored the whole taxonomy to match where we are headed: applied AI, with WordPress and AI-for-WordPress as the bridge.

What I liked was the discipline of it. The agent read the live taxonomy and wrote a report before it touched anything. Then a dry-run plan I signed off on. Then it executed, with a rollback snapshot and a check before every delete. Twenty categories became seven plus an archive, fifty-five tags became four, and every old URL got a redirect so nothing breaks.

The model did the busywork. We kept the decisions and the paper trail. That is the kind of AI-for-WordPress work we are doing now.

I wrote up the whole thing, including the near-verbatim session so you can see exactly how it ran, over on Insights: Refactoring fifteen years of blog taxonomy in one session.

If you run a WordPress site that has drifted from where your business actually is, and you are a San Francisco Bay Area company, feel free to reach out. We are here for you.