Search “AI related posts plugin” and you will find a lot of marketing language. Words like smart, intelligent, and advanced algorithm show up everywhere. Real AI, the kind that reads what a post actually means instead of counting shared words, shows up almost nowhere. We checked six popular WordPress related-posts plugins against their own documentation to find out which is which.
The short answer
None of the six plugins below use AI or semantic matching. All six match posts by counting shared tags, categories, or keywords. One plugin that did use real AI for this exact job used to exist. It is gone now, and nothing has replaced it.
| Plugin | How it actually matches posts | Real AI or semantic matching? |
|---|---|---|
| YARPP | Weighted score from MySQL fulltext search on title and body, plus tags and categories | No |
| Contextual Related Posts | MySQL FULLTEXT index on title and content | No |
| Jetpack Related Posts | Elasticsearch index of content, categories, and tags, run on Automattic’s servers | No, based on every public source |
| Related Posts by Taxonomy | Counts how many taxonomy terms two posts share | No |
| Inline Related Posts | Shared category or tag | No |
| MonsterInsights | Not a relatedness feature. Its “Popular Posts” tool ranks by pageviews, comments, or shares | Not applicable |
YARPP (Yet Another Related Posts Plugin)
YARPP has been around since the early days of WordPress plugins, and its own listing explains the method plainly: a match score built from your post’s title, body text, tags, categories, and custom taxonomies, each with its own weight you can adjust. Under the hood, that text score comes from MySQL’s fulltext search, a real and useful technique. It is keyword overlap, not an understanding of what a post is about. You also set a relevance threshold, so YARPP only shows posts that clear a minimum score.
Contextual Related Posts
Contextual Related Posts calls its matching an “intelligent algorithm” that scans your title and content. The plugin’s own technical notes say what actually happens: on activation, it builds MySQL FULLTEXT indexes and searches those. The paid version adds an “Advanced Algorithm” that lets you manually reweight title, content, excerpt, and taxonomy. Every option here is a setting a person configures, not something a model learns.
Jetpack Related Posts
Jetpack’s own support page says the feature scans your content, categories, and tags, then runs everything through Elasticsearch on WordPress.com’s servers. Elasticsearch is a full-text search and relevance engine, the same category of tool search engines use, not a semantic or embedding-based system. Automattic does not publish every detail of its internal scoring, so we cannot rule out some statistical layer we cannot see. Based on everything Automattic has published, this is full-text and taxonomy matching at cloud scale, not AI matching.
Related Posts by Taxonomy
This one is the clearest case. The plugin’s own tagline says it directly: posts sharing the most taxonomy terms show up first. No content analysis. No scoring model. Just a count of shared categories, tags, or custom taxonomy terms across one or more post types.
Inline Related Posts
Inline Related Posts spends most of its marketing on where related posts appear rather than how they are chosen. Its own placement technology detects paragraph breaks so a related-post box does not land mid-sentence, which is a genuinely useful trick. For the matching itself, the plugin’s listing and independent write-ups agree it works from shared categories or tags you select, the same taxonomy approach as the others on this list.
MonsterInsights
MonsterInsights does not have a feature called related posts. Its closest equivalent, Popular Posts, solves a different problem: it shows what is popular sitewide, ranked by pageviews from Google Analytics, comment count, or social shares. That is useful for a “readers also liked” sidebar, but it has nothing to do with whether a post relates to the one someone is currently reading. Worth knowing if you searched for AI-matched related posts and landed on this plugin by mistake.
Why so few plugins actually do this
It isn’t that semantic matching is a secret technique nobody’s thought of. It’s that shipping it inside a plugin most people install for free changes what the plugin has to be.
Tag and category matching runs entirely inside WordPress, on data you already have, with no external dependency. Real semantic matching needs an embedding for every post, generated by calling out to a machine learning model, either self-hosted (expensive, and most shared WordPress hosting can’t run one) or through an API like Voyage AI or OpenAI (which costs real money per call, however small). That’s an ongoing cost and an external dependency a plugin author has to support forever, not a one-time feature to build and forget. For a free plugin with no revenue behind it, that math often just doesn’t work, which is a plausible, honest reason this stays rare, even without knowing the specific finances of any one plugin that’s tried and stopped.
Marketing words to actually check before trusting them
Every plugin above uses at least one of these phrases somewhere in its own marketing, and none of them, on their own, mean AI. “Intelligent algorithm” almost always means a scoring formula with adjustable weights, not a trained model, so it’s worth asking what the algorithm is actually scoring. “Smart matching” is the same claim in different words. “Advanced algorithm” is usually just the paid-tier version of the same fulltext or taxonomy scoring, with more weights you can adjust, not a different underlying technique.
“Understands your content” is the one worth the closest look, because it’s the phrase that sounds most like a real AI claim. A genuine semantic system understands content in the sense of representing its meaning as a vector. A keyword system “understands” content only in the sense that it can find matching words, which is a much smaller claim wearing similar-sounding language.
The one thing that’s actually hard to fake: a named embedding model, a named vector comparison method, or an explicit mention of an external AI API being called. If a plugin’s documentation can’t point to any of those three things, it isn’t doing semantic matching, whatever the homepage copy says.
Does semantic matching actually matter for your site
Not always, and it’s worth being honest about when the free, keyword-based version is already enough. If your posts are tagged consistently and specifically, tag overlap alone gets you most of the way to a good related-posts list, because the vocabulary problem semantic matching solves, two posts about the same thing using completely different words, mostly doesn’t happen when tagging is disciplined in the first place.
Where it earns its cost is a large, older archive with inconsistent tagging, or a site where the same topic legitimately gets covered under different vocabulary across years of posts. That’s exactly the case keyword and taxonomy matching structurally can’t solve, no matter how well you tune the weights, because there’s no shared word or term for it to find in the first place.
The plugin that actually tried AI
One plugin in this space did build real machine learning matching. Bibblio’s Related Posts for WordPress plugin used AI to read your content and learn from how readers actually clicked, to the point where its own reviewers noted you no longer needed to tag posts carefully for it to work.
That plugin is discontinued. We do not know the exact reason. What we do know is that nobody has picked up where it left off. As of today, we are not aware of another actively maintained WordPress plugin, outside our own, that uses real AI to match related posts.
How Ariadne is different
Ariadne’s Free tier works the same way as the six plugins above: tags, titles, and text overlap, computed entirely on your own server. Pro and Pro+ add a second signal that works in a completely different way. Every post gets a semantic embedding from Voyage AI, a vector that represents what the post means, not just which words appear in it. Ariadne blends that signal with its ordinary tag, title, and text matching, so a post still gets a sensible related-posts list even before it has an embedding of its own.
None of this runs while a visitor is loading your page. Every related-posts list is computed ahead of time and read from a cache. The Voyage AI call is opt-in, uses your own API key, and Ariadne works fully without it if you never turn it on.
What this means if you are choosing a plugin
If a related-posts plugin’s page uses the words smart, intelligent, or advanced algorithm, check what that phrase is actually describing before you take it at face value. In every case above, it described fulltext search or shared taxonomy terms. Both are real, useful techniques. Neither is AI. If semantic matching is what you actually want, here is the honest state of things today. One real AI-based option shipped in this space, and it got discontinued. A small number of current plugins, ours included, are building it fresh.