Richardson Cloud, Rebuilt: Faster, Cheaper, AI-Driven

Welcome back. Richardson Cloud has a new home, and a new way of being built.

For the last few years this blog ran on WordPress, which meant an App Service plan and a MySQL database quietly billing every month for a site that is, at its core, a collection of articles. That’s a lot of moving parts, and cost, for static content.

So it has been rebuilt as a static site on Azure Static Web Apps. The result:

• Negligible hosting cost. The Free tier covers a blog comfortably.
• Faster pages. Pre-rendered HTML with no server round-trip.
• Same analytics discipline. Lightweight, first-party page metrics captured to Azure Table Storage, plus Google Search Console for the search side.

That’s the what. The part I care about is the how.

Built the way I think we should build now

I didn’t just move the content. I rebuilt the delivery pipeline to practice what I preach about AI-driven DevOps, and I want to be specific about what that means, because “I used AI to build my blog” usually implies a division of labor that isn’t quite what happened here.

• Everything is code. The infrastructure is defined in Bicep, the site in a typed content model, the pipeline in a single workflow. Nothing important lives in a portal where it can’t be reviewed.
• Tokenless, automated delivery. Push to a branch and GitHub Actions runs infra → build → deploy → test, authenticating to Azure with OIDC, no secrets sitting in a vault waiting to leak. Infra only re-deploys when infra actually changed.
• AI in the loop, not just in the copy. I built this with AI as a pair — scaffolding, reviewing, and hardening the pipeline — and the same assistant now collaborates with me like a content manager: it drafts articles for me based on the content roadmap and outlines we’ve built together (including diagrams and images, in markup form), pushes them to dev for me to review or edit, and ships to prod once I give final approval. That’s the real unlock: AI compressing the boring parts of the loop so the judgment calls get more of my attention, not less.

None of that is exotic. It’s IaC, CI/CD, and a real division of labor with AI, and the point is that it’s consistent and repeatable, from a one-person blog up to an enterprise platform. Cloud is a model, not a place. DevOps is a discipline, not a tool. And AI is most useful when it’s doing the typing while you keep doing the thinking.

What’s next

More writing, and more of it: practical, opinionated pieces on Azure architecture, AI development and governance, AI-driven DevOps, and building the teams that ship. Same honest take you’ve come to expect, published on a cadence again, starting with a closer look at exactly how that AI-driven pipeline works and what it’s allowed to decide without me.

How are you wiring AI into your delivery pipeline, or are you still deciding where it earns its place? I’d love to hear it. More soon.