There was a blog post published this week that’s been rattling around in my head for a couple of days. He asked a fair question about the WordPress plugin business. When AI can read your docs and write your features, what is a license actually paying for?
His answer was/is lock-in. Move your logic behind a cloud API so an AI agent can’t read it, can’t copy it, can’t replicate it. Make the software harder to inspect and you make the business harder to clone. I think Casey is sharp. I think he’s pointed at the right problem. But I think the answer he reached for is the part we should be nervous about, because it solves the wrong fear.
There’s a question sitting one layer underneath his, and almost nobody in WordPress is saying it out loud.
Yet another plugin
For years, the WordPress plugin business let you get away with thin. You could wrap one hook in a settings page, give it a name, and sell it. We even had a convention for it. “Yet Another [Whatever] Plugin.” It started as a joke. Then it stopped being one. A good chunk of what shipped was somebody’s weekend project with a price tag stapled on. Don’t believe me, here it is: Yet Another Related Posts Plugin.
That worked for exactly one reason. Copying was expensive.
Not the code. The code was usually trivial. What was expensive was the distance between seeing your idea and shipping your idea. To copy you, somebody had to know PHP, know the hooks, stand up an environment, and burn their nights doing it. So most people who could have copied you never did, because the friction cost more than the payoff was worth. And so we told ourselves a comfortable story. We said the code was the moat many wouldn’t dare cross.
It never was. We just couldn’t see that yet, because the lights were off.
AI didn’t raise the bar
Here’s the thing people are getting backwards right now. Everyone in the WordPress plugin business is saying AI raised the bar. That customers expect more. That you can’t ship thin anymore. That the standard went up. But the bar didn’t move. The bar was always there.
AI turned the lights on.
The floor of real value in WordPress has always existed. Some products were standing well above it and some were standing below it and getting paid anyway, because the cost of copying was high enough to hide the difference. When a customer couldn’t open a code editor and ask an AI “is this actually doing anything, or could you rebuild it in an afternoon,” the thin product and the deep product looked the same from the outside.
Now the customer can ask. And so can your competitor. The gap was always real. It’s just clearly visible from space now and everyone has 20/20 vision.
Why Gravity Forms isn’t sweating
Look at the products nobody’s worried about. Gravity Forms. Easy Digital Downloads. ACF.
Could an AI invent something like them? Sure. But there are ten thousand moving parts in those plugins that the person “building their own version” will never see, because you can’t copy what you can’t perceive. They won’t replicate the hooks they don’t know exist. They won’t handle the edge cases they’ve never hit. They’ll build the demo and miss the decade.
And notice why those products are safe. It is not because AI raised the bar and they cleared it this year. They were always above the bar. AI didn’t make them more uncopyable in 2026 than they were in 2024. The depth was always there. What changed is that AI finally exposed the cheap stuff standing next to them.
So depth is only visible to people who can already see depth. That cuts both ways, and the second way is the uncomfortable one.
The lie the plugin business tells itself
When people talk about AI and copying, the plugin business reaches for the same nightmare. Some nobody with Claude is going to materialize your product overnight and undercut you.
That fear is popular for a reason. It’s flattering.
It puts the threat outside you. It makes the danger a tool, not a person. It means if you get beaten, it wasn’t really a fair fight, it was the robot’s fault, tough break. So you get to keep believing you were good and the world just got unfair.
But it’s also wrong.
A novice with an AI doesn’t become your competitor. AI removes their excuse for not starting. It does not hand them the one thing that actually beats you, which is judgment. The person who can take your idea and ship something better was never going to be stopped by the difficulty of writing the code. They were always capable. They just hadn’t bothered yet.
The fear the plugin business won’t say out loud
So here’s the real one. Say it plainly.
The threat isn’t the copy. Someone will copy you. And the honest answer to whether someone could build your thing with AI is yes, it always was. They don’t even need your codebase to start. The day you reveal what you’re doing, somebody more capable than you can start doing it too.
The real fear is that someone outclasses you, someone who was always able to, and AI just removed the friction that used to keep you from finding out.
They won’t beat you because they used AI. They’ll beat you because they’re more experienced, and AI collapsed the time between them seeing your idea and them executing on it. All they ever needed was the idea. And you handed it to them in your launch announcement.
So the friction was never a moat. It was a hiding place. And the same thing that’s exposing thin products to customers is exposing comfortable builders to better builders. This is the lock-in conversation from the other side: you don’t get to hide anymore, and neither does your code.
The gut check
This is the moment to be honest with yourself, because the easy move is to dodge.
It’s tempting to read all this, decide your product is one of the deep ones, and move on. Most people will. They’ll picture the imaginary novice, feel safe, and close the tab.
But you’re in a bubble. So am I. We look at our own work, see all the thought that went into it, and assume we’re ahead. Meanwhile there’s an enormous world of people building right now, with these same tools, doing more than you’d guess from inside your own little corner. Some of them are better than you. And some of them are already working on the thing you think you own.
So the question isn’t “can my product be copied.” It’s “if someone more experienced than me pointed themselves at this tomorrow, what do I actually have that they can’t shortcut?”
And if the only honest answer is “a head start,” the head start is already gone.
What the WordPress plugin business actually sells
Here’s where I land, and it’s the opposite of Casey’s lock-in.
The thing that survives a smarter person with a faster tool is the thing made of time. Not the code. The code is free now. What’s scarce is everything code can’t be: being genuinely useful, doing exactly what you said you’d do, and refusing to trap people in something they impulse-bought and can’t get out of.
A competitor can copy your feature in a weekend. But they cannot retroactively have shown up for two years. They can out-engineer you. But they cannot have already earned the trust. That’s the one kind of depth AI can’t accelerate, because it isn’t made of code. It’s made of time, and time doesn’t compress.
Casey wants to keep customers by hiding knowledge so they can’t leave. I’d rather keep them by giving them so much that they don’t want to. Both produce retention. But one makes a hostage, and one makes a regular. That’s the WordPress plugin business I want to be in, and I know which end of it I’d want to be standing on.
I’ll go first.
Because I’m not preaching this from safety. I build in the open. I give the base version of my work away for free and build the paid layer on top of it, in public, where anyone can watch and anyone can copy. So if somebody beats me to it, I’m not going to complain. I’ll have learned where my floor actually was. I’ve written before about who actually wins in the AI era, and it was never the person with the best-hidden code.
So that’s the question Casey’s post left me with, and the one I’d hand to anyone in the WordPress plugin business right now.
You’re not really afraid of the clone. You’re afraid the clone will prove the thing was thin.
If your product can survive being fully understood, you don’t need a moat. But if it can’t, no cloud API is going to save you. It’ll just buy you twelve months and teach your customers you were keeping something from them the whole time.
The lights are on now. Go look at what you built.
Leave a Reply