WordPress AI Plugin Lock-In Is the Second Thing Nobody Is Talking About

WordPress AI plugin lock-in is the second thing nobody is talking about. The first was WordPress AI token cost. I wrote about that last week. The short version: there’s a measurable gap between how developers prompt AI models and how regular users do, and that gap shows up as a real cost difference most plugins aren’t accounting for. If you missed it, go read that first. This post builds on it.

After that post landed on Reddit, one comment cut through the noise. The commenter pointed out that AI features aren’t just adding a new cost layer to WordPress plugins. They’re making lock-in significantly worse. They’re right. Plugin lock-in has existed in this ecosystem forever. AI doesn’t just continue that tradition. It deepens it in a way that’s harder to see coming and harder to recover from.

WordPress AI Plugin Lock-In Is Different From Anything We’ve Seen Before

Traditional plugin lock-in is painful but recoverable. You switch page builders and you lose your layouts. You migrate form plugins and you rebuild your forms. And it goes deeper than layouts and fields. If your contact form was connected to Stripe or PayPal, you’re also losing sales conversion data, form entries, and all the meta information attached to them. By the time you’re ready to make that switch, you’re either fed up with the plugin or you’re switching to cut costs. You can justify the loss. It hurts, but you can get past it.

The thing you lose is configuration. Structure and settings that live in the database and can, at least in theory, be exported, mapped, and moved somewhere else.

AI lock-in is different. What you lose isn’t configuration. It’s workflow. And workflow lives in your head.

Spend three months using Yoast’s AI suggestions to write meta descriptions and you’ve learned how that specific model responds to your content. You know what it nails, what it misses, how to steer it toward what you actually want. That knowledge doesn’t transfer when you switch tools. It evaporates. You’re not rebuilding your site structure. You’re rebuilding how you think. That’s a fundamentally different kind of switching cost, and most people evaluating these plugins right now have no idea they’re signing up for it.

Divi’s Flat-Fee Model Is the Most Aggressive Lock-In Play in the Ecosystem

Divi’s pitch is straightforward: pay one price, get unlimited AI generation. No token counting. No billing surprises. After my post about token costs, that pitch sounds almost protective.

We’ve seen this playbook before. Image optimization plugins ran the same play. Unlimited compression for one price, until the costs became unsustainable and the model quietly shifted to per-gigabyte pricing or tiered plans. Early adopters got the good deal. Everyone else got repriced once the habit was formed and switching felt too painful.

That’s exactly where Divi’s flat-fee AI model is headed. The unlimited generation offer gets you in the door, builds your workflow around their tooling, and then in a year or two the pricing changes. By then you’re dependent enough that leaving feels harder than paying more.

But there’s a deeper problem than the pricing bait. That model severs your relationship with the underlying AI entirely. You don’t have a Claude account or a OpenAI account. You have a Divi account. The AI is a feature of the plugin, not a tool you own access to. Which means if you ever leave Divi, for any reason, you don’t just lose a page builder. You lose your entire AI-assisted creative workflow and you have nothing to carry with you.

Bring Your Own Key plugins are at least honest about the relationship. The plugin is a wrapper. Your API key is yours. Wrappers can be replaced. But even BYOK doesn’t solve the workflow lock-in problem. You still build muscle memory around how a specific plugin uses your API key. The key survives a migration. The workflow doesn’t.

I’m Not Just Saying This. I’m Operating This Way.

The Reddit commenter asked whether keeping AI out of the dashboard entirely is the only future-proof approach. For me, right now, the answer is yes. And I want to be clear that I’m not just saying that. I’m doing it.

This post came out of my own AI process. It started with a conversation in Claude Desktop about the Reddit comment. That conversation surfaced factors I hadn’t fully considered, which shaped what I actually wanted to say. I wrote the post the way I write everything, using my text to speech process to get my thoughts down. Then I used Claude Code to handle the technical side, setting the Yoast SEO fields, fixing the warnings, publishing to my site. No AI writing plugin. No dashboard widget. My words, my process, my tools working together the way I’ve set them up.

That’s the point. The AI relationship lives in my Claude account, not inside any plugin. I can change anything about my WordPress setup and not lose a single part of how I work.

The WP Abilities API Is the Right Direction. Here’s Why.

Think about what it would take to replicate this workflow inside the dashboard with plugins. You’d need an AI writing plugin for drafting. A Yoast AI add-on for SEO. Probably an AI research or content brief plugin. A grammar and readability tool on top of that. You’re looking at four or five plugins minimum, each with their own AI integration, their own model, their own pricing, and their own lock-in.

The WordPress community has spent years preaching against plugin bloat. Too many plugins slow your site, expand your security surface, create update conflicts, and introduce failure points you didn’t sign up for. Now we’re about to stack four or five AI plugins on top of an already heavy install and call it innovation.

And here’s the part that makes it worse. Those four or five plugins aren’t going to stay independent. We’ve seen this pattern play out in WordPress for over a decade. A category gets hot, someone builds the leading plugin, and a bigger company acquires it and folds it into their ecosystem. Yoast bought Thirsty Affiliates. WP Engine bought ACF. Automattic absorbed WooCommerce and built Jetpack into a bundle that touches half your site. The acquisitions are coming. The AI writing plugin you’re depending on today gets bought by Yoast next year. Your SEO AI gets folded into their premium tier. Suddenly you’re not locked into five plugins. You’re locked into one company that controls your entire AI-assisted workflow.

The WP Abilities API cuts through all of that. When the AI lives outside WordPress and communicates through a structured, documented API, the acquisitions become irrelevant. It doesn’t matter who buys what. The interface is stable. Your workflow is portable. You’re not dependent on any plugin staying independent, staying affordable, or staying focused on your needs.

That might change as the ecosystem matures. But as of today, one tool talking to WordPress through a clean API is a better architecture than five plugins each trying to own a piece of your AI workflow.

What Plugin Developers Should Be Asking Before They Build Anything

If you’re a plugin developer thinking about adding AI features, here’s the question worth sitting with before you write a single line: does this feature make the user more capable, or more dependent?

There’s a version of AI in WordPress that genuinely helps people do things they couldn’t do before. And there’s a version that trains them to need the plugin to do things they could learn to do themselves. The second version feels like a feature. It’s actually a trap. For the user, and eventually for the developer when the trust runs out.

• Keep the API relationship theirs. BYOK isn’t optional. The user’s key stays with the user.
• Make content fully exportable. If AI touched it, the user should be able to take it anywhere.
• Disclose the model. What’s powering this feature? What version? Users building workflows around your plugin deserve to know.
• Don’t build workflows that only work inside your plugin. That’s not a feature. That’s a subscription disguised as one.

None of this is technically difficult. It’s a choice. The plugins that make these choices will build trust. The ones that don’t are making a short-term bet that lock-in will keep users from leaving. That bet has a history in WordPress, and it doesn’t tend to end well for the users caught in it.

There’s a Third Layer Under WordPress AI Plugin Lock-In

Token costs were the first thing nobody was talking about. WordPress AI plugin lock-in is the second. The third is the assumption running underneath both of them. That WordPress users need AI to do things they’re not capable of doing themselves. That’s the one that bothers me most, and it deserves its own conversation.

This is part of a series. Read the next post. The WordPress Agentic Web.