WordPress and the intersection of AI

When I talk about WordPress, people here in San Francisco usually give me a puzzled look.

“Did you know WordPress powers 43% of all websites and growing?”

I tell them.

“Yes, it’s absolutely absurd. Who uses WordPress anymore? Framer and WebFlow are going to crush WordPress”

Meanwhile, Framer and WebFlow collectively have less than 3% of the market share for CMS platforms, and WordPress has over 60%. Sorry, Framer and WebFlow are not going to overtake WordPress anytime soon.

While I’ve moved past the idea that WordPress will be the major vehicle for the AI future (that idea was something I believed briefly when I first realized the enormity of WordPress, the power of open source communities, and the staying power/switching costs of software), I still deeply admire WordPress, and think there’s an interesting story to be told about WordPress and AI.

The original thought was that AI could penetrate into WordPress’s ecosystem and be a large surface area for introducing AI into the day to day happenings of the world. While this will likely still happen, it will happen everywhere. Not just WordPress.

The release of NLWeb from Microsoft is a great example of how quickly the world is going to uptake WordPress – within days of it opening, multiple people (myself included) began building plugins to WordPress to turn it into NLWeb servers (allowing AI Agents to search and query existing WordPress sites, as well as regular users having an AI chatbot that could ask any question of the WordPress site, and have it surface the relevant content).

So no, the reason I keep talking about WordPress isn’t because I think it will be a large focal point of the AI future. It’s because it’s a useful and inspiring analogy about what the future of open source could hold for the age of AI agents.

I’m inspired by Matt Mullenweg. Despite the negative press & controversy that’s surrounded him, he’s the one who introduced me to the idea and concept of the power of open source software.

He visited my hometown Salt Lake City, and spoke about open source at the Silicone Slopes Summit in 2023. I was in the audience, and I remember being shocked when he told us how his little project he started when he was 19, WordPress, took over the world.

He was talking a different language that I wasn’t used to when it came to startups: personal freedoms, the right to ownership, democratizing publishing. This was lofty language that resonated with a founder in his early 20s.

He challenged the audience to consider building open source software, and that maybe they could build something that had as large of an impact on the world as WordPress.

By making WordPress extremely easy to download, install, and deploy, he was giving a gift to anybody who wanted to start a blog and publish their ideas. This helped spark a revolution, and later helped small business owners publish critical information about their business, that allowed them to show up on placed like Google.

But that was 2004, and my obsession began with WordPress in 2024. In 20 years, how did WordPress become this massive behemoth that has an $800B world-wide ecosystem?

And yet, when I first tried WordPress, even though I have a degree in Computer Science and have been using computers daily since 4 or 5 years old, it was confusing and clunky.

And yet, it’s still so dominant.

This is where I understand people’s misunderstandings: WordPress? Really? But Framer, WebFlow, Wix, SquareSpace, they’re all so much easier!

There’s no way WordPress will be around for long.

And this is the question I’ve obsessed over for all of 2024. And I think I’ve found the answer. WordPress isn’t going away. I’ve boiled it down to three things:

1. Because it’s open source
2. It has a huge community
3. It has a large and open 3rd party marketplace that creates huge network effects and causes massive switching costs.

And all 3 of these things, actually give WordPress a competitive advantage over all of it’s competitors.

When Microsoft announced NLWeb, you had multiple indie hackers/developers jumping at the opportunity, to introduce the protocol into WordPress’s ecosystem. I’m sure this was happening before the top executives at Framer, Webflow, Wix, or Squarespace had even realized what NLWeb actually is. And this is why WordPress actually is a better, more durable tool than any of it’s competitors.

It’s infinitely customizable and flexible, and this means there is more deep innovation happening on it’s platform than anywhere else. This means that new innovative ideas can actually be rolled out faster to WordPress users than it’s competitor’s users. For example, our AI coding agent, LlamaPress, has a slick integration with WordPress that means that WordPress has a better AI native building & editing experience than any other page builder out there currently. WordPress literally has: prompt to website creation, with all the flexibility that comes with it, before it’s competitors.

Because some random hacker in San Francisco decided to code up a 3rd party integration with it. Matt Mullenweg and I have never talked 1 on 1. But in a way, I’m an employee and asset of Automattic. All the 3rd party WordPress developers of the world are. Because in their sum, Automattic has an army across the globe of people working non-stop 24/7, providing free time, energy, innovation, and intellect, to helping increase WordPress’s capabilities through 3rd party extensions and plugins.

And that’s why open source, deep communities, and 3rd party extensibility holds deep competitive advantages over closed source proprietary systems.

This is also why I think LangChain and LangGraph are going to be big winners in the new AI agentic world. I’m betting big on LangChain and LangGraph.

But that does beg the question: is WordPress the right analogy to compare against LangChain and LangGraph? And also, where would open weight models such as Llama and Deepseek fit into this?

I will write a future article going deeper into this idea. But briefly, here’s how I view things.

1. Open LLMs are analogous to the Linux operating system. They open up a base-layer of possibility where you’re not locked into a proprietary vendor.
2. LangChain and LangGraph are analogous to other open source software, such as PHP, Apache Web Server, and MySQL. LAMP is the “wrappers/orchestrators” of the operating system, just as LangChain/LangGraph are the “wrappers/orchestrators” of the LLM.
3. What is the equivalent of WordPress? WordPress is essentially a “wrapper” of the LAMP stack, right?

The equivalent of WordPress is LlamaPress. Will this bear into fruition? Will LlamaPress one day support a $800B world-wide ecosystem of builders? The future is uncertain, there is no clear answer yet.

But I know that 2024 looked a lot like 2004. And in 2004, WordPress was a tiny community, of raving, excited contributors, who all probably didn’t conceive of taking over the world. And that’s what LlamaPress currently is.

We are a wrapper of LangGraph, LangChain, and LLMs, so that non-technical users can publish their own micro web-apps and micro AI agents. They can self-host LlamaPress on their own servers at home, and they can build a software and AI system that they truly own.

What an exciting time to be alive as a builder.