I built this site years ago intending to write regularly. I never did. Now after using it as a testbed to experiment with agentic AI it’s back – refreshed and ready for some new prose.
Years ago I threw up this web site with some intentions of writing and sharing my ideas but I never got back to regular publishing. I enjoyed crafting the design and the technical creation. Writing was not really my focus at the time and I had no real idea of what I wanted to communicate. I still have no firm idea – at least this time I want to write more and share my thoughts on various topics that interest me. Hopefully in time this site can reflect me in some way.
I have created many sites over the years in professional web development, for other people, and also for myself. My first experience with the Web wasn’t even online but in a computer class in 1994 with a Notepad editor, an INDEX.HTM file, and Netscape or Mosaic browser. Since then the web has made many advancements in HTML as well as adding JavaScript and CSS. Today we have a dynamic application platform that builds on these technologies, far beyond the rich documents and linking features from the beginning years of HTML. As the complexity of the web grew so did the way people published their content online. People went from uploading hand-edited pages to their service provider’s site, to blog publishing platforms; today most of it happens on social media. Tomorrow, I believe that as AI improves the ease of creating fun and pleasing web content we might find more personal websites being self-published again.
Like many people in the IT industry I had been using AI for the last few years and tried early agentic AI programming in late 2025. It was OK but not reliable. Something changed in early 2026. There was some kind of model tipping point and suddenly agentic AI was good and has only gotten better. I needed to experiment with agentic coding and I wanted a real project, so I decided to play around with my neglected web site. At first I was just experimenting, then after some time I thought: why not just update it? It was fun to have ideas then to see them implemented quickly and working after a few minutes. It sparked my creativity as ideas flowed without the friction of working out the technical implementation.
My old site used Jekyll and I wanted to keep the static site generator but I also wanted something a bit more modern. I had already read a bit about Astro but I hadn’t used it before and I wanted to learn more about it. Using my old site as a reference I had the AI agent set up Astro and quickly started giving it requirements about the look I wanted, the fonts, the style, and what to bring over or leave behind. I also wanted to bring a bit of fun into the site and thought of many ambient effects I could apply to different articles. This was done quickly by the agent and beyond my expectations although I still had to guide things along and iterate a bit when things didn’t go quite right.
I worked by asking questions about anything I wanted to understand. Then once I had settled on the approach I wanted to take I’d let the AI build it. I checked the results. Asked for more edits. If that experiment didn’t work or I didn’t like something then I’d just discard that piece of work. I checked in the changes I liked regularly so I could easily go back to what was working, much like when I work on code myself. The process is the same as not using AI agents: research, experiment with what works, iterate until the result is good. There are many other ways AI can help too – skills that critique your design, check for security issues, review your writing, or help you set up and deploy your site.
One trade-off from using AI to build your personal web site is that you may not learn anything new. In professional settings learning a new technology falls into two categories: learning the minimum you need to get something done, or becoming a knowledge expert in that technology (since you use it all the time). Using AI requires the same choice. There are many ways to use AI, from helping you learn while doing the development yourself all the way to just doing things for you, or perhaps a middle ground where you understand what’s being done at a high level but the details are left to the AI. How you use AI may depend on circumstance – do you want to learn something or do you just want it done?
Another trade-off is that you can be too confident in the working result and fail to check for other issues. Design was a big one – new changes often broke consistency or introduced stray new elements. The same could be said for code quality, so it is always good to review everything after a lot of changes.
I hope AI can help bring back some of the fun of the early web and that people use it to make more interesting, playful, and creative personal sites. What started as a work experiment ended with me enjoying it so much that I just committed and gave my site a modern refresh. It’s 2026, not 1994, but I feel the same sense of imaginative opportunity. And text matters as much as it ever did, only now AI can take care of the code while I take care of the writing.
