Why MDX Makes Internal Linking Obvious (Even for AI)
Internal linking is important for SEO and UX, but finding good links is boring manual work. MDX changes that by giving both humans and AI actual structure to reason about.
Internal linking is one of those things everyone agrees is important.
It helps SEO. It helps users. It helps content age better.
And yet, it's usually painful.
Not because people don't understand why they should do it, but because finding good internal links is boring, manual work. You scroll, you skim, you search, you open tabs, and you still feel like you're guessing.
This is where MDX quietly changes the game.
The real problem with internal linking in WordPress
In a typical WordPress setup, your content lives as a big HTML blob.
It might come from:
- a classic editor
- Gutenberg blocks
- a page builder
- shortcodes
- block comments mixed into markup
Even if the content looks fine in the editor, structurally it's messy.
For a human, skimming that HTML is annoying. For AI, it's worse.
You can paste the HTML into a prompt, but now the AI has to:
- ignore layout markup
- ignore wrapper divs
- ignore inline styles
- infer structure from noise
So most "AI internal linking" tools end up doing keyword matching instead of understanding intent.
That's not an AI problem. That's a structure problem.
This is part of a broader pattern we explore in WordPress is a Chain of Compromises.
MDX is closer to an outline than a blob
An MDX post isn't just content. It's structure.
You have:
- real headings
- clear sections
- predictable ordering
- named components instead of anonymous markup
Even before rendering, an MDX file already tells a story.
You can glance at it and understand:
- what the post is about
- what the main sections are
- where certain ideas live
That matters a lot.
Because internal linking is not about keywords. It's about context.
Why AI works better with MDX
When AI looks at MDX, it doesn't need to "read" everything.
It can:
- jump between headings
- understand section boundaries
- reason about intent
- suggest links based on meaning, not proximity
Instead of "find matching words", it can answer questions like:
- this section explains X, what other post expands on X?
- this paragraph introduces Y, do we already have a guide on Y?
And it can do that without being fed hundreds of lines of HTML.
That's the key difference.
AI doesn't need magic prompts. It needs content it can reason about.
This is why it's time to talk about headless WordPress in the context of AI.
A very different workflow
With MDX, internal linking becomes something you can do after writing.
A realistic workflow looks like this:
- Write the post
- Let it sit
- Ask AI: "What internal links make sense here?"
- Review and apply
No scanning HTML. No plugins guessing intent. No manual search across the site.
Because the content is structured, the suggestions are actually useful.
This isn't about AI hype
This isn't about "AI writing your content".
It's about removing busywork.
The same pattern shows up everywhere:
- components instead of templates
- files instead of database blobs
- predictable structure instead of clever abstractions
AI benefits from that, but so do humans.
MDX just happens to be a very clear example of this principle.
Why this matters in a WordPress context
WordPress is still the CMS most clients know.
They're comfortable with it. They trust it. They expect it.
But the way we deliver content around WordPress doesn't have to be stuck in the past.
Using MDX alongside WordPress lets you:
- keep WordPress for content management
- keep MDX for content structure
- get the best of both worlds
And crucially, it keeps your content future-friendly.
If you ever change CMSs later, your content doesn't collapse into a migration nightmare. It's already clean. This is part of what makes migrating WordPress to Astro a sensible long-term play.
Structure beats cleverness
Internal linking is a great example of a broader idea.
AI doesn't struggle because WordPress is old. It struggles because WordPress content is often unstructured.
Give AI structure, and suddenly a lot of "hard problems" become boring.
And boring, in this case, is good.