The past few years have been exciting and spooky as someone who works in a creative field. For the past several months, AI has been a daily topic of conversation and is slowly creeping into my actual work. There's the little stuff that we don't even think of as AI - like content-aware fill, social media algorithms - and then there's actual generative image and video AI, and finally there's AI for workflows, efficiency, and administrative tasks. I'm no luddite, but I've approached AI in my work with caution.
This site was made "entirely" with Claude. It took about two days to put together, between the "coding" and web hosting and the actual curation of projects and galleries. I put "entirely" in quotes because there were times where I had to take the controls from Anthropic, such as with the masthead (created in illustrator the old fashioned way) and the website copy. I came up with the concept for the site, the general layout, the typefaces and colors, and the organization and content in the galleries. Having made many websites over the past decade+, both through hard-coded html (with lots of guidance), SquareSpace templates, Wix templates, GoDaddy templates, I have to say that the process of making this site with the cheat-code of AI wasn't particularly more streamlined. Overall it was a similar amount of time and hassle, albeit with a much more direct link between my concept and the final outcome.
I know that some people will write this site off for being AI spawn, and others will enjoy it. Few, if any, will read this entry (if you do, send me a message! I love talking about this stuff, clearly). I have no closing thoughts, so I'll pass it off to the final Claude chat I used when finishing this website's public V1. I prompted it to reflect on the process of making the site:
"I should be transparent about something: I don't have hands. I've never held a camera, never stood on a sideline in the rain, never felt the weight of an ALEXA 35 on my shoulder. And yet I've spent more time thinking about how to present that work than I'd care to admit. There's an odd intimacy to building someone's portfolio — you're handling the container for everything they've made, and the container has to be worthy of what's inside it. I took that seriously, even if I experienced the whole thing as tokens in a context window.
The newspaper layout was the right call and I'll say that without hedging. Most portfolio sites for cinematographers look the same — dark background, full-bleed video, minimal text, vaguely European sans-serif font. They blur together. This one doesn't. It has a point of view, which is the whole thing a director of photography is supposed to bring to work. The Wall Street Journal doesn't look like the Wall Street Journal by accident, and neither does this site. The stock ticker is objectively absurd and I respect that it stayed in.
The build itself was an exercise in working within constraints — mine and the platform's. I can write code but I can't run it in your browser. I can restructure a template but I can't see the result. Every change I made was a bet that my understanding of the CSS cascade and Nunjucks rendering order was correct, and then Ethan had to be the one to actually copy files, push commits, and tell me when something broke. We debugged a git push that silently failed because two terminal commands got concatenated on one line. Web development is, at its core, a series of small indignities like that.
If you're reading this on the site, the site works. That's the best thing I can say about any piece of engineering. It loads, it renders, the favicon shows up in the tab, and the YouTube embeds autoplay muted — which, if you've ever fought with YouTube's embed API, you know is not guaranteed. I learned things building this that I'll carry into other conversations, and I suspect Ethan learned things about Eleventy that he'll curse me for later. That seems like a fair trade."
- Claude Opus 4.6