![\"CMS](\"https://andrewembler.com/_astro/image.BB1ohIrI.png\")
Photo by Iswanto Arif on Unsplash
I have to come clean.
\nIn the first post in this series, I wrote that the CMS I use now is “one I built myself, for exactly one user.”
\nThe “one user” part is true. The “built myself” part needs a bit of an asterisk. Almost every line of code behind this site and its editor was written by Claude.
\nIt’s no secret that this personal blog fell into atrophy over the years. Work and life overtook whatever meager ambitions I had to be a blogger or a thought leader. When it came time to reconstitute this thing, I had a pretty strong vision, but I still didn’t have any more time than I used to.
\nClaude has nothing but time. The funny part? With Claude involved it turns out I didn’t need much time after all:
\nSo: a couple of days, from nothing to a designed, deployed blog with its own authoring app. The site and the editor are about 10,000 lines of source between them. I didn’t type most of them.
\nSomewhere in there, I posted a thought that pretty much sums up the experience: watching a machine churn through your job in seconds really does a number on the ol’ imposter syndrome.
\nMostly, I decided things. It’s actually not that surprising: as the core team leader of Concrete CMS I have been making decisions around software for many years, especially around code I didn’t write. That turns out to be most of the job.
\nI decided the content should be plain Markdown files in folders. I decided each post’s colors should come from its hero photo. I decided the first version wasn’t right, even though it worked, and started over. I decided we also needed a lightweight CMS “Studio” that could help me edit even when a Markdown editor wasn’t handy.
\nAnd I tested. A lot. The Safari bugs alone (a white flash between pages, a stutter after the photo animation) took a few rounds of “no, still broken, here’s what I see” before they were really fixed. None of that is typing. (Which is a pity, because that’s still a skill I’m good at.)
\nWhen I start a post, I can have it draft a title and description. It suggests alt text for photos. It can tighten, expand or rephrase a paragraph. And there’s a little robot in the corner of every post, which I named BeBot, that I can chat with about the draft. It proposes edits as diffs. It’s very good at fixing typos and grammar mistakes that I make when I’m frantically churning through a post.
\nAnd while we’re airing all of our dirtiest laundry: I like this. I like it a lot. It turns out that I really hate the imposing nature of a blank page. If I can have an LLM give me the world’s clunkiest first draft of something, I’ll take it. The earliest, firstest of passes on this topic came from Claude; I gave it the title and one sentence about what the post was about, and it read the first post, dug through the git history, and was off to the races. But what you’re reading now bears little resemblance to it.
\nI’m not going to pretend I have this figured out. I lead an open source project whose whole reason to exist is helping people build websites, and I just built one in a day without really writing it.
\nWhat I know is that the site is better than anything I would have built by hand, because I wouldn’t have built anything by hand. I had years to do it and didn’t. The friction was the problem, and that friction is gone.
", "date_published": "2026-10-03T03:25:00.000Z", "tags": [ "Rebuilding this site" ] }, { "id": "https://andrewembler.com/posts/im-optimistic-about-the-blazers-anyway/", "url": "https://andrewembler.com/posts/im-optimistic-about-the-blazers-anyway/", "title": "I'm optimistic about the Blazers anyway", "summary": "Tom Dundon has spent his first six months cutting staff, skipping playoff T-shirts and eyeing the exits. I'm not gonna let that bring me down.", "image": "https://andrewembler.com/_astro/hero.D1oYUhM5.jpg", "content_html": "I can’t help it. I’m excited about this Blazers season.
\nNot “excited, but.” Just excited. How can you not be, when we get to experience “Dame Time” again? And that’s not even the wildest part about the Blazers’ off-season!
\nThe Blazers’ sale to Tom Dundon closed this March. Six months in, here are the highlights so far:
\nNone of that is illegal. But a basketball team isn’t only a business. It’s one of the few things a whole city shares. When the owner treats free T-shirts as waste, he’s telling you how he sees the people wearing them.
\nYes, a lot of this is just how a certain kind of businessman runs things. But I don’t love it.
\nHere’s the thing, though. Last season the Blazers went 42-40, got through the play-in, and took a game from the Spurs in San Antonio before losing the series in five. That was with an unexpected coaching change at the start of the season, Damian Lillard on the sidelines the entire time, rehabbing a torn Achilles, and Scoot Henderson gone for most of the year.
\nThe people who made that happen are mostly still here:
\nThen add the backcourt. Lillard is back, healthy, in a Blazers jersey again. And the team traded Jerami Grant and Kris Murray to Memphis for Ja Morant, which is not a sentence I expected to type a year ago.
\nYes, I am bummed about the loss of Shaedon Sharpe. Yes, the Blazers’ guard rotations are still mysterious and loaded with small defensive liabilities. This is the team’s third head coach in less than twelve months.
\nYahoo Sports projects them at 34-48, which would be a step backward. I’ve watched a lot of Blazers teams that were safe bets to be mediocre. This one isn’t safe. The Nori hire could be great. Portland has a knack for rehabilitating faded stars. The defense might be a problem. The games won’t be boring.
\nPortland fans are a persnickety bunch. We’re gonna find something to have a problem with. Am I worried about the existential future of the Blazers at the hands of an owner whose relationship with the region is complex at best? Of course. But I can’t do anything about that. What I can do is appreciate the greatest Blazer of all time coming back for one more go-round. Who doesn’t want to see that?
", "date_published": "2026-10-03T02:02:56.000Z" }, { "id": "https://andrewembler.com/posts/moving-out-of-the-house-i-built/", "url": "https://andrewembler.com/posts/moving-out-of-the-house-i-built/", "title": "Moving out of the house I built", "summary": "I co-founded Concrete CMS. I'm still its core team leader. But this blog doesn't run on it anymore. Does that feel strange? You bet.", "image": "https://andrewembler.com/_astro/hero.1w0LN6EF.jpg", "content_html": "Photo by Josh Hild on Unsplash
This site doesn’t run on Concrete CMS anymore.
\nThat’s a strange sentence for me to write. When we first open sourced Concrete in 2008, I couldn’t wait to use it on my own website. This blog has run on it in one form or another ever since. (At least for a while: after 2017, I mostly let the place atrophy. For crying out loud – I never even put it behind SSL!)
\nIt made sense: at its height Concrete CMS (then named concrete5 – yep, that’s right, with a lowercase “c”) powered hundreds of thousands of websites of many sizes, across many verticals. We needed to eat our own dog food! When something broke in Concrete, I often found out here first.
\nBut today, the site is a folder of Markdown files, built into static HTML and served from Cloudflare. The CMS I write in is one I built myself, for exactly one user.
\nIt’s bittersweet.
\nI want to get this out of the way first: if you’re running a content-heavy site with a bunch of people who need to manage it, Concrete CMS is still the tool I’d reach for. Lots of pages, lots of editors, different people allowed to touch different things: that’s exactly the job it was built to do. The original vision of “see a typo, fix a typo” right from the page? It still delivers on that, and we are privileged to still have a group of extremely passionate, knowledgeable, prolific and talented developers working with our open source project.
\nBut my blog has none of those problems. It has one author. I’m a developer, and happiest in a text editor. Most of what Concrete is good at, I wasn’t using. At PortlandLabs we work hard to ensure our tech stack is kept up to date, but I definitely did not apply the same rigor to my personal DigitalOcean droplet: as it became annoying to keep everything updated, I let that friction keep me from posting, and it wasn’t long before my blog was a ghost town.
\nEvery post is a Markdown file, with its images in the same folder. The whole site lives in git. It sounds small, but it changes what owning the site feels like. I can search the whole archive with BBEdit. And if every tool I use today goes away, I’ve still got a folder of text files that pretty much any computer can open. And metadata can just live in the post alongside the content:
\n---\ntitle: \"Moving out of the house I built\"\ndate: 2026-10-02\ndek: \"I co-founded Concrete CMS. I'm still its core team leader. But this blog doesn't run on it anymore. Does that feel strange? You bet.\"\nheroAlt: \"Brown wooden house surrounded by green trees during daytime\"\nheroCredit:\n name: \"Josh Hild\"\n url: \"https://unsplash.com/@joshhild?utm_source=andrewembler_studio&utm_mediu>\n source: \"Unsplash\"\n sourceUrl: \"https://unsplash.com/photos/brown-wooden-house-surrounded-by-gree>\nseries: \"rebuilding-this-site\"\nseriesPart: 1\ndraft: trueA static site has no PHP to update or database to back up. Static HTML will always be the fastest, most secure way to host a website. We employ talented DevOps folks to do this; it was never something I was interested in or particularly skilled at. I’m glad I can now delegate that problem to Cloudflare!
\nThis last part is the toughest part to admit: my blog had grown stale, and I didn’t really want to improve it. I had ideas about a design that adapted to the colors present in imagery, but I couldn’t justify the amount of time it would take to achieve it. Yes, you can (and always have been able to) build anything in Concrete, but at some point the idea of clicking on images and micro managing content just for a post struck me as just too fiddly. For a personal blog, it was too much management and not enough content.
\nThis is the part I’m most excited about.
\nEvery general-purpose CMS has to make compromises. Concrete makes them on purpose, and I’ve spent a long time defending those choices. A CMS has to work for a church, for a university, for a law firm, and for someone’s band. Concrete has always been powerful and flexible, because it had to be.
\nBut mine doesn’t. It does what I want and nothing else. No, it’s definitely not built in a vacuum: this site runs on the great work done by Astro, alongside a custom CMS named… you know what? It doesn’t even have a name! I think we’re finally post-product, here.
\nWhen I want a new feature, I add it. When I stop using one, I delete it. There are no packages to update, and no users dependent on something I must decide I no longer want to support. The only person I have to argue with is me (and I usually lose.)
\nFor years, I’ve been building software meant to work for everyone. It’s an interesting feeling to build something that only has to work for me.
\nI’ve spent 20 years telling people that a good CMS lets you focus on content instead of infrastructure. And here I am, having built my own infrastructure so I could focus on content.
\nThe right tool depends on who’s holding it. For an organization with editors, a CMS like Concrete takes away a pile of problems. For one developer with a blog, some of the problems were the CMS itself. Sometimes all you need is a folder full of files.
\nIf you’re reading this, you can safely assume that I succeeded!
\nIf you’re still using Concrete CMS – thank you! Don’t forget to update to 9.5.5 next week! Concrete powered this site for a long time, and I’m grateful for that. It just doesn’t need to anymore.
", "date_published": "2026-10-02T23:54:00.000Z", "tags": [ "Rebuilding this site" ] } ] }