I am almost in total agreement with you... let me explain.

I have only played with free AI coders, ChatGPT, Perplexity, Claude, Google AI Studio and Telex by Automattic, to create WP plugins. None of them managed to code a complete plugin with one try. Some (mainly AI Studio and Claude added extra functionalities I did not ask, which were excellent so I kept them). Some if not all of them were also making syntax errors, which is unacceptable. Logic errors are understandable, but not syntax errors.

$600 is a lot of money, but compared to how much it was going to cost you if you had hired humans to code those things, not to mention the time they would need, $600 is not even a bargain, it's straight up robbery.

Now the hard question from me to all... If AI is so good and so fast and very cheap, how come we don't see tens of thousands of new CMS to overthrow or at least compete with WP, E-commerce competitors to Woo, Shopify etc... ? We are not going to see new social media platforms as the trick there is not the software, but the expensive infrastructure, bandwidth, servers, hard disk, RAM, etc...

I predict in the future for all he current AI tools to become better, and hopefully with competition to become even cheaper, and to see a whole new type of tools to run locally on our computers, machine learning algorithms to make and train our own AI models as we please. And that's when individual software developers will be valuable again, it will be who can train their own in-house AI better, instead of everyone having access to the same AI tools.

I also predict that sooner rather than later the EU will pass a whole new set of garbage laws to disrupt most of those advances, for the European citizens, so do as much as you can now...

Thanks Nick, very good thinking as always.

There is a world of difference between free and paid AI coding. Claude Opus 4.6 churns out complete web applications without a single issue. Dozens of models, controllers and views.

No new WordPress-alike systems? That's an easy one: we don't need that anymore. Yesterday, Claude Sonnet spit this out: https://pietercooreman.github.io/ASPPY. My prompt? One line: https://claude.ai/chat/1d408f12-60d3-4c70-af7b-32f97ceac863

I guess in the near future, we're gonna see dozens of competitors for Claude, ChatGPT and Gemini. And AI will get cheaper. And we'll see new operating systems, new browsers and new industry-specific AI-driven solutions. That's where the money is now: industry-specific AI-driven software and robots that replace humans.

You can already use AI on a local machine with tools like OpenCode and Claude Code. And you can instruct them to use specific libraries that sit in a given local folder. That's how I developed ASPPY. The Python code of my transpiler simply sits in the same directory and is available to OpenCode to go inspect in case of difficult implementations. That seems to work fine.

And yep, the EU will try to protect its citizens against e-tech companies stealing their data. I am brainstorming about a product that keeps all my images, videos, mails, documents, on a local cloud system at home. In total I have 10TB of disk space in network drives, PC's, USB drives, that are not used. And I pay Google 120 EUR/year to store 1TB of images and videos? It's a waste of money. We need to move away from the cloud and use the TB's of local storage we have at home. And indeed, train our own AI models on them. I'm on it

I think the answer to why aren't we seeing new CMS or social media platforms is twofold. Time and first mover advantage.

Lots of users complain about platforms but imagine someone taking a niche subject and creating their own platform. Say, a sub Reddit about woodworking. Run it on a home server and offer a better experience. People will desert Reddit. So, fragmentation but over time.

With CMS I suspect it's currently a question of "if it ain't broke don't fix it" but... again, I think that's a time issue. The big factor there is that fragmentation works in favour of WP. Nobody is likely to gather enough traction to move a million developers and a gazillion sites no matter how good it is

For now, and probably the next 2 years, WordPress will not lose many users to vibe coding tools to create websites. Only when that becomes virtually free, users will start to migrate I guess.

The main problems of WordPress are performance, security, learning curve and bloat. Vibe coding tools fix that by creating smaller applications that only focus on what a user really needs and uses, keeping page load times far below 100ms without any caching plugin. And with the right prompts, vibe coding tools can create much more appealing and modern websites in seconds, than a WP theme like Twenty Twenty Five is able to.

You literally need at least 5 plugins to create a website like https://pietercooreman.github.io/ASPPY/ in WP. Maybe more. And plugins can get left behind, outdated, or show security and performance problems. Again, not a problem with vibe coding. You throw-in your website in a vibe coding tool and have it fixed or updated with just spoken language. And it works. This is a different methodology. We're living in a new era. WordPress is solid and well thought of. But WP is also boring and hard to get used to, even for developers and designers. Vibe coding tools are smarter and automate everything. Once you're used to that, you will never return to manually editing a WP site. The WP-way is simply too slow, too complicated and it requires a lot of babysitting and patching. I'm not gonna miss WP... I can tell....

Eventually, I was also involved with Pieter due to his enthusiasm for AI.
And using Claude (free), I tried my hand at creating a small system for uploading files and their attributes, managing categories and subcategories, and storing files in the file system of the corresponding categories and subcategories (creation if non-existent and deletion when the file is deleted or moved from category to category).
The directive given to Claude was to use ASP Classic, Access as the database, and datatables for data visualization.
In the end, considering the limitations of Claude Free, I managed, or rather, Claude managed to build the mini file management system.
My observations:
The problem is that Claude (I'm referring to the free version) has little understanding of ASP Classic for building a complete system from scratch, even though it's simple in its functionality.
He probably knows ASP syntax, even though he used DISTINT (not supported by Access) in a SQL query to Access.
He had problems with datatables and jQuery (for example, with JavaScript calls when the script hadn't yet been loaded).
And many other bugs that forced Claude to generate 73 zip files with the code to download before getting the working version.
Claude probably isn't very knowledgeable about ASP, and I think that's because he has little to track down and learn from the web today, given that ASP Classic is no longer used by anyone, and therefore online resources are increasingly scarce.
My idea is that the logic Claude applies to ASP Classic derives from a sort of logic taken and applied to other more widely used languages ​​today, such as PHP and Python.
That said, and in any case, with considerable effort on my part to debug the code provided by Claude, the result was achieved.
Of course, you need to have some knowledge of web programming and what all this entails. If I hadn't had this knowledge, I would have thrown the first zip file created by Claude in the trash, as well as the second, the third... up to 73, which is the valid zip.
I tried to formulate my requests to Claude in a valid and technical manner, and not as if I were a programming novice. Despite this, the result for me wasn't having the system ready in a few seconds or minutes, but rather several hours of interaction with Claude, trying to make him understand what I wanted and, above all, getting him to fix his bugs.

Attached is an image of what I asked Claude.

Hi Sergio, thanks for sharing.

You’d better use a tool like Opencode.ai. You install it on your local PC, point it to a directory (you’d better create a new one), create a new project in Opencode, and start a session. Big Pickle is a free and decent AI at OpenAI.

Next, it's all about the prompt. For ASPPY prompting, I use a prompt builder:
https://pietercooreman.github.io/ASPPY/prompt_builder.html
.
Besides the project description, it also includes dozens of important rules, do's and don'ts, sample code, etc.

Now that I'm thinking about it, it would be a good idea (and it would take only a few minutes) to create such a prompt builder for AI-based Classic ASP/VBScript development.

And finally, there is a huge difference between free models like Sonnet/Haiku and Opus. The same goes for GPT Nano (free) and paid GPT 5.X Codex or any Pro AI coding model. A HUGE difference... also for Classic ASP. There are still millions of lines of ASP/VBScript code floating around on places like GitHub. QuickerSite alone has 300,000, of which quite a few are very well structured and mature.

I see it as a life goal to — by the time I die — have generated a few more million lines of very well structured and MVC-enabled ASP/VBScript code. With ASPPY I'm focusing on that, and it works. And you're right: AI coding models base themselves on existing programming flow logic and often guess. The good news is: we can help by generating well-structured code.

New runtimes for Classic ASP like ASPPY and AxonASP might help as well. However, they will also add to the confusion, because they both support much better file handling (ZIP, PDF, images, binaries in general), whereas in Classic ASP you needed quirky workarounds or third-party components.

Actually, we really need ONE new ASP/VBScript runtime that quickly becomes the new standard for ASP coding. I'm talking with Lucas, the developer of AxonASP. We already agreed on making our systems 100% compatible. More to come...

Thanks Pieter for the advice and the tools to use.
You and Lucas are certainly doing a great job trying to keep ASP Classic alive.
However, from what I've seen, and if I understand correctly, to use ASPPY or AxonASP, you need a dedicated server. Classic hosting isn't enough, and this is the type of service that most users (and perhaps even businesses/corporations) use.
If that's the case, I think having a dedicated server could become an obstacle to the spread of ASPPY and AxonASP.

We're not just trying to keep ASP alive. We both extended it with advanced file handling (jpg, zip, pdf, etc), database compatibilities, JSON support. Lucas even added 60+ new functions borrowed from PHP. We're making sure ASP/VBScript will outlive the platform they were built for (Windows/IIS).

Nowadays you can pull up Linux servers for as low as 5 USD/month, or less. Both AxonASP and ASPPY run faster on Linux than Classic ASP/VBScript ever did in IIS, thanks to compiled languages in the back (Go for AxonASP, Python for ASPPY).
I run several ASPPY sites using IIS and its rewriting engine. Without any dependency on ASP.dll or the Windows Scripting Host. One small web.config does the trick: https://github.com/PieterCooreman/ASPPY/wiki/Replacing-Classic-ASP-VBScript-by-ASPPY-in-IIS-%E2%80%90-Easy-way

Be aware VBScript is scheduled to be removed from Windows OS. Hosting companies will therefore soon stop offering Classic ASP hosting. That's the reason behind our efforts. To keep legacy ASP applications alive, and offer a more robust and modern framework for them. Platform independent.

Yes, Pieter, I know full well that vbscript will no longer be supported sooner or later.
And I wish you with all my heart success in this new venture you're embarking on.
My observation about dedicated servers, which can influence the success or otherwise of ASPPY and AxonASP, isn't limited to their cost alone, but above all to their system management.
To manage a dedicated server (businesses and companies aside), users must also have system knowledge of the operating system they intend to use. Whether it's Linux or Windows doesn't matter, but they absolutely must have this knowledge.
It's true that nowadays, with a single command line, once your dedicated server is configured, you can install what you need (with Composer, for example), but you have to pray that everything goes well and the installation is successful. If that doesn't happen, without system knowledge, you won't get out of this situation, because at that point you have to understand why "composer" didn't do what it was supposed to do.
I know full well what problems can arise from managing a dedicated server because in the company where I worked, I switched from the old Sun Microsystems Solaris operating system (Unix), then to Linux, then to Windows Server and desktop, and finally to web programming, first with PHP (ASP Classic didn't exist yet), and then both PHP and ASP Classic.
This is the gist of my observation about dedicated servers. If this isn't a problem for most users in the future, so be it!!! ...and then long live ASP Classic again with its new extensions...and I'll be happy for that.

Sergio, check this out https://youtu.be/2Ft3m0RRnCc

You'll never use free AI again.