I have built a product which uses AI to built Astro sites. LLM builds the sites in steps and make sure that they get 100/100 scores in pagespeed insights. These sites are served with a CDN. You can edit the sites with LLM interface, or use markdown editor to edit sites, or edit texts directly on a dashboard. These sites are static. There is no vendor lock in. If you want to migrate and manage yourself, just go to cloudflare or github pages. These sites cost 0 eur to run, and they always score better in all benchmarks compared to sites that are built on top of a separate CMS server.
I know WordPress is going nowehere and if there is some special backend functionality, that is needed. But 95% of web does not need it.
A static site is always cheaper, and the bottleneck has always been that editing code is indimidating. Therefore, AI actually resolves a big problem here, and this is going to alter the future of platforms like WP.
What do you use to edit texts on a dashboard or cms?
My wife needed a website. I’m not much of a frontend coder but I looked at the price of squarespace for a year and decided to go that route with AI.
Website is Astro. I easily update sophisticated designs with Claude or Gemini cli.
Hosted for free on cloudflare, it’s super fast. Any git update deploys to the website in a minute.
Got an hosted email form with astro action.
The only thing left is integrating a CMS. I was thinking of keystatic but it’s not compatible with Astro 6 yet. That’s the issue with vibe coding a stack you don’t know as much, without realizing it picked a version of astro that was so new that some tools didnt quite integrate with it yet.
Ditto. Almost all of my internal tools are built in the same fashion and deployed on cloudflare itself.
It is so much liberating to have no overload of CMS and the bloated code they come with.
A CMS is for content authors, not developers. AI making it easier to build or re-platform sites doesn’t change the need for an admin UI that non-technical people can actually use, or the need for access control, governance, and approval workflows around how content gets changed.
If anything, I’d expect more CMS work to occur as the cost of building, migrating, and redesigning sites keeps dropping.
There is indeed not always a need for WordPress. I have been using ProcessWire (1) for over a decade. Open-source, zero dependencies, no-nonsense CMS — and when it comes time to build a new website, I go back to it even in 2026, because you make it once and it works for 10 years and counting.
Cloudflare is just jealous that most of their customers are actually running WordPress, but this is not something they will be able to solve with AI hype.
+1 for Processwire! I’ve mentioned it here a few times over the years and nobody seems to have ever heard of it! I’ve got a few sites well past 10 years now still happily chugging away on it! Basically zero issues with it, ever. It’s still my go to for all sorts of projects - installs in a few seconds, loads of really useful functionality out the box, easy API, beautifully flexible for all sorts of projects and a great community and ecosystem around it as well!
This looks extensible - how is the marketplace for finding people who can work on it? Wordpress is so common because it's so common you can always find someone to hack it together.
I expect the bigger risk to dynamic database-backed CMS platforms right now is that AI assistance makes static site generator tools run against a version controlled repository of content less intimidating for most users... and static sites are cheaper to run (especially in this era of badly coded scrapers flooding the internet) and much less likely to fall vulnerable to security problems.
I expect we'll see a further wave of CMS interfaces which provide a nicer editing experience on top of flat files stored in Git.
Maybe the strategic move for platforms like WordPress (and maybe Django too! The Django admin remains a very popular CMS platform) is to invest more in separation of admin editing from serving, such that there's an obvious path to edit your content in the CMS but deploy it as static files.
My own blog uses the Django admin and serves the site via Django (albeit behind a 15m Cloudflare cache to handle traffic spikes) but I have a scheduled GitHub Action that backs up the content to a Git repository: https://github.com/simonw/simonwillisonblog-backup - it's not much of a stretch from that to having the Git repository feed content to a static site generator.
There's Tina CMS for editing files in Git, but honestly editing flat files is probably the least interesting or complicated part of an enterprise CMS, and IMO there's rarely a good reason to interact with files directly versus a database that publishes files.
I ran git-based blogs for years and have gone back to CMS. The instant preview and the instant publishing really make it a lot more pleasant to work with it. With Git, my read-eval-print loop so to say was a minute which is just too long. Fixing a typo then takes 2 minutes.
> and static sites are cheaper to run (especially in this era of badly coded scrapers flooding the internet)
Is that really such a problem for the average Joe? I'm running multiple blogs via a Rust CMS [1] on the cheapest Hetzner server, and have had no problems with the scrapers or load or anything. Have also gotten to the HN front page without issues talking about that you shouldn't put a site behind Cloudflare since most don't need it [2]. Now of course, for businesses or something who depend on the service to be online, it's different. But I'm talking about regular Joe's blog here.
Depends on the company level, on my line of business, what companies care about are headless CMS, with AI workflows, and oriented towards MACH.
Ah, and all of them have partnerships with Vercel, and possibly Netlify.
Sitecore, Contentful, Sanity, Storybrook,...
If anything, they killed the need for backend skills, you get a ready made SaaS, program interactions with AI, and if anything requires backend like logic, it is taken care by Vercel or Netlify functions.
I wouldn’t drop WordPress for Astro - but I did for Hugo. Never been happier.
My perspective comes from enterprise: we use(d) a marketing agency to run two websites. A few months ago I discovered our team was spending 30+ minutes just to publish a blog post written by a product manager. Everything was built on Elementor blocks. Articles pasted from Word kept breaking styles. 20+ plugins creating a security nightmare.
With AI assistance, we migrated to Hugo in three days. 800+ pages. 15 reusable components. Zero plugin chaos. Permissions handled at the git level. A simple HTML form to upload images and paste articles for less technical people, most were fine with markdown already. GitHub Actions for cleanup, validation, and spellchecking. Attack surface minimized. Performance improved drastically.
I’ll stand behind this: most people don’t need a bloated CMS. They need clarity on what they want to achieve, a solid process, and software that turns that process into a system.
I agree with the sentiment against blindly jumping into the AI wagon from CMS, but the author seems mixing that from the migration from CMS into markdown content + static site generation.
To me the latter is a legit move and much cleaner architecture for most sites. And the issue of editing code, or really just markdown files, seems to be a solvable UI problem with good editors like Obsidian, or something similar but more tailored for website building.
I never really understood the argument, you can design the frontend with AI anyway and then use WordPress as the CMS for clients. Clients want to be able to log in, update CPTs, edit a calendar, post pictures, things like that. Surely the idea isn't to vibe-code an admin panel/cms from scratch for every project?
For client projects I tend to go with KirbyCMS. Easy license, great functionality, very easy to setup and configure for the client side users. And way less bloated. I actually really dig it and nearly exclusively use it for my projects.
But other tools are great as well, like ProcessWire (named above).
I’m not sure that I buy all the points made. I can imagine an AI centric CMS where the technical interface (implement this site on MySQL, host it there, use Next.js, etc) is distinct from the content interface (change store hours) or even the design (change the background).
I have used Wordpress a lot (too much) and came to the view that for most websites it is just overkill. So I built https://harcstack.org and vowed to write all my new sites in actual code.
HTMX to the rescue since you can write server side code in a sensible way and still have quite a dymanic UX.
The same way coding agents don’t replace the need for an IDE, content generation needs to support arbitrary human-to-agent handoffs, where the human can say “this is the wrong direction, I sketched this change of what I want it to look like, see how it’s different and apply that pattern.”
And, in the broadest sense, that human interface is a CMS; the agent is just another editor, albeit one that happens to read and write raw data rather than using a WYSIWIG (or similar) editor.
But it needs a better headless capability. Most separate front ends appear to be grafted on relying on plugins. Which doesn’t make a whole lot of sense.
> Joost de Valk, founder of Yoast SEO, wrote about how he migrated his personal blog from WordPress to Astro, the hottest new JavaScript framework in town, and suggested the blasphemous idea that not all sites need a CMS (he’s since migrated again to EmDash, which I talk about later)
That's a weird thing to read. (Not criticism for the author or the article)
> Migrated his personal blog
Is that a thing worth mentioning? I did that over a decade ago.
> Astro, the hottest new JavaScript framework in town
I thought it's 2026 now, not early 2010s. People still do that?
> the blasphemous idea that not all sites need a CMS
Yoast SEO is a huge paid plugin for Wordpress, so the founder not using Wordpress is at least moderately interesting, in a "cobbler buys shoes at walmart" kind of way.
I have built a product which uses AI to built Astro sites. LLM builds the sites in steps and make sure that they get 100/100 scores in pagespeed insights. These sites are served with a CDN. You can edit the sites with LLM interface, or use markdown editor to edit sites, or edit texts directly on a dashboard. These sites are static. There is no vendor lock in. If you want to migrate and manage yourself, just go to cloudflare or github pages. These sites cost 0 eur to run, and they always score better in all benchmarks compared to sites that are built on top of a separate CMS server.
I know WordPress is going nowehere and if there is some special backend functionality, that is needed. But 95% of web does not need it.
A static site is always cheaper, and the bottleneck has always been that editing code is indimidating. Therefore, AI actually resolves a big problem here, and this is going to alter the future of platforms like WP.
What do you use to edit texts on a dashboard or cms?
My wife needed a website. I’m not much of a frontend coder but I looked at the price of squarespace for a year and decided to go that route with AI.
Website is Astro. I easily update sophisticated designs with Claude or Gemini cli.
Hosted for free on cloudflare, it’s super fast. Any git update deploys to the website in a minute.
Got an hosted email form with astro action.
The only thing left is integrating a CMS. I was thinking of keystatic but it’s not compatible with Astro 6 yet. That’s the issue with vibe coding a stack you don’t know as much, without realizing it picked a version of astro that was so new that some tools didnt quite integrate with it yet.
Ditto. Almost all of my internal tools are built in the same fashion and deployed on cloudflare itself. It is so much liberating to have no overload of CMS and the bloated code they come with.
A CMS is for content authors, not developers. AI making it easier to build or re-platform sites doesn’t change the need for an admin UI that non-technical people can actually use, or the need for access control, governance, and approval workflows around how content gets changed.
If anything, I’d expect more CMS work to occur as the cost of building, migrating, and redesigning sites keeps dropping.
There is indeed not always a need for WordPress. I have been using ProcessWire (1) for over a decade. Open-source, zero dependencies, no-nonsense CMS — and when it comes time to build a new website, I go back to it even in 2026, because you make it once and it works for 10 years and counting.
Cloudflare is just jealous that most of their customers are actually running WordPress, but this is not something they will be able to solve with AI hype.
1. https://github.com/processwire/processwire
+1 for Processwire! I’ve mentioned it here a few times over the years and nobody seems to have ever heard of it! I’ve got a few sites well past 10 years now still happily chugging away on it! Basically zero issues with it, ever. It’s still my go to for all sorts of projects - installs in a few seconds, loads of really useful functionality out the box, easy API, beautifully flexible for all sorts of projects and a great community and ecosystem around it as well!
This looks extensible - how is the marketplace for finding people who can work on it? Wordpress is so common because it's so common you can always find someone to hack it together.
A client of mine is using ProcessWire as his site was done by the guys behind process wire. Big shoutout. To the system and the guys.
Great system as far as I am concerned - even if I tend to use KirbyCMS for most of my projects.
I expect the bigger risk to dynamic database-backed CMS platforms right now is that AI assistance makes static site generator tools run against a version controlled repository of content less intimidating for most users... and static sites are cheaper to run (especially in this era of badly coded scrapers flooding the internet) and much less likely to fall vulnerable to security problems.
I expect we'll see a further wave of CMS interfaces which provide a nicer editing experience on top of flat files stored in Git.
Maybe the strategic move for platforms like WordPress (and maybe Django too! The Django admin remains a very popular CMS platform) is to invest more in separation of admin editing from serving, such that there's an obvious path to edit your content in the CMS but deploy it as static files.
My own blog uses the Django admin and serves the site via Django (albeit behind a 15m Cloudflare cache to handle traffic spikes) but I have a scheduled GitHub Action that backs up the content to a Git repository: https://github.com/simonw/simonwillisonblog-backup - it's not much of a stretch from that to having the Git repository feed content to a static site generator.
There's Tina CMS for editing files in Git, but honestly editing flat files is probably the least interesting or complicated part of an enterprise CMS, and IMO there's rarely a good reason to interact with files directly versus a database that publishes files.
I ran git-based blogs for years and have gone back to CMS. The instant preview and the instant publishing really make it a lot more pleasant to work with it. With Git, my read-eval-print loop so to say was a minute which is just too long. Fixing a typo then takes 2 minutes.
> and static sites are cheaper to run (especially in this era of badly coded scrapers flooding the internet)
Is that really such a problem for the average Joe? I'm running multiple blogs via a Rust CMS [1] on the cheapest Hetzner server, and have had no problems with the scrapers or load or anything. Have also gotten to the HN front page without issues talking about that you shouldn't put a site behind Cloudflare since most don't need it [2]. Now of course, for businesses or something who depend on the service to be online, it's different. But I'm talking about regular Joe's blog here.
[1]: https://github.com/rikhuijzer/fx
[2]: https://news.ycombinator.com/item?id=45965060
Depends on the company level, on my line of business, what companies care about are headless CMS, with AI workflows, and oriented towards MACH.
Ah, and all of them have partnerships with Vercel, and possibly Netlify.
Sitecore, Contentful, Sanity, Storybrook,...
If anything, they killed the need for backend skills, you get a ready made SaaS, program interactions with AI, and if anything requires backend like logic, it is taken care by Vercel or Netlify functions.
I wouldn’t drop WordPress for Astro - but I did for Hugo. Never been happier.
My perspective comes from enterprise: we use(d) a marketing agency to run two websites. A few months ago I discovered our team was spending 30+ minutes just to publish a blog post written by a product manager. Everything was built on Elementor blocks. Articles pasted from Word kept breaking styles. 20+ plugins creating a security nightmare.
With AI assistance, we migrated to Hugo in three days. 800+ pages. 15 reusable components. Zero plugin chaos. Permissions handled at the git level. A simple HTML form to upload images and paste articles for less technical people, most were fine with markdown already. GitHub Actions for cleanup, validation, and spellchecking. Attack surface minimized. Performance improved drastically.
I’ll stand behind this: most people don’t need a bloated CMS. They need clarity on what they want to achieve, a solid process, and software that turns that process into a system.
I agree with the sentiment against blindly jumping into the AI wagon from CMS, but the author seems mixing that from the migration from CMS into markdown content + static site generation.
To me the latter is a legit move and much cleaner architecture for most sites. And the issue of editing code, or really just markdown files, seems to be a solvable UI problem with good editors like Obsidian, or something similar but more tailored for website building.
I never really understood the argument, you can design the frontend with AI anyway and then use WordPress as the CMS for clients. Clients want to be able to log in, update CPTs, edit a calendar, post pictures, things like that. Surely the idea isn't to vibe-code an admin panel/cms from scratch for every project?
For client projects I tend to go with KirbyCMS. Easy license, great functionality, very easy to setup and configure for the client side users. And way less bloated. I actually really dig it and nearly exclusively use it for my projects.
But other tools are great as well, like ProcessWire (named above).
I’m not sure that I buy all the points made. I can imagine an AI centric CMS where the technical interface (implement this site on MySQL, host it there, use Next.js, etc) is distinct from the content interface (change store hours) or even the design (change the background).
I have used Wordpress a lot (too much) and came to the view that for most websites it is just overkill. So I built https://harcstack.org and vowed to write all my new sites in actual code.
HTMX to the rescue since you can write server side code in a sensible way and still have quite a dymanic UX.
I'm a fan of Concrete CMS. No plans to switch away, and yes I am working with AI.
CMS is pretty much alive, even if most of them are now headless, oriented towards MACH deployments and AI based workflows.
I didn't know about MACH, interesting.
I made a thing [1] a few months ago because I wanted a lightweight expression of this.
[1] https://github.com/bootstrapital/flatcontent
> But that idea is old enough to drink
Nice turn of a phrase! I was surprised it was a GoogleNope except for you, op.
The same way coding agents don’t replace the need for an IDE, content generation needs to support arbitrary human-to-agent handoffs, where the human can say “this is the wrong direction, I sketched this change of what I want it to look like, see how it’s different and apply that pattern.”
And, in the broadest sense, that human interface is a CMS; the agent is just another editor, albeit one that happens to read and write raw data rather than using a WYSIWIG (or similar) editor.
> coding agents don’t replace the need for an IDE
Depending on who you talk to, they may not agree. (I am not in this camp but I am certainly aware of people who are.)
Wordpress, like SQL, is probably immortal.
But it needs a better headless capability. Most separate front ends appear to be grafted on relying on plugins. Which doesn’t make a whole lot of sense.
> Joost de Valk, founder of Yoast SEO, wrote about how he migrated his personal blog from WordPress to Astro, the hottest new JavaScript framework in town, and suggested the blasphemous idea that not all sites need a CMS (he’s since migrated again to EmDash, which I talk about later)
That's a weird thing to read. (Not criticism for the author or the article)
> Migrated his personal blog
Is that a thing worth mentioning? I did that over a decade ago.
> Astro, the hottest new JavaScript framework in town
I thought it's 2026 now, not early 2010s. People still do that?
> the blasphemous idea that not all sites need a CMS
Is it? People still haven't accepted this?
>founder of Yoast SEO,
>wrote about how he migrated his personal blog from WordPress to Astro
>he’s since migrated again to EmDash
Do you need to know anything more about this guy? If that's one of the articles sources, I think you can ignore anything it says.
Yoast SEO is a huge paid plugin for Wordpress, so the founder not using Wordpress is at least moderately interesting, in a "cobbler buys shoes at walmart" kind of way.
> Is that a thing worth mentioning? I did that over a decade ago.
Is that a thing worth mentioning? I abandoned my personal blog a decade ago.
/s but only so slightly.