Wondering how PurioChat understands a visitor’s question instead of hunting for exact words? The answer is embeddings. Here’s what they are.

The short version

An embedding is a numerical representation of your content. PurioChat reads a post or listing and turns its meaning into a long list of numbers (a vector) — a “fingerprint” of what the content is about, not just the words it contains.

Because meaning is stored as numbers, the AI can measure how close two pieces of text are in meaning, even when they share no words at all.

Why this beats keyword matching

Traditional WordPress search looks for the exact words a visitor typed. If those words don’t appear, nothing is found. Embeddings capture meaning instead, so similar ideas land near each other.

A visitor searching for a “cozy place to stay” will match a listing described as “comfortable accommodation”, even with no shared keywords. The meanings are similar, so the embeddings are close together, and PurioChat surfaces the right result.

Tip: Picture every piece of content on a giant map of meaning. Similar things sit near each other, and a visitor’s question lands on the same map so PurioChat can find the closest content.

Where embeddings come from: training

Your content has no embeddings until you create them — that’s what training does. When you run training, PurioChat sends each item to your AI provider, gets back its embedding, and stores it for searches and chatbot answers.

You do this from PurioChat → Data Training. On the free plugin you can train your Posts (on any theme) and, on the Listeo theme, your Listings. Until you’ve trained some content, PurioChat has nothing to compare against and shows a “Please train the AI first” notice instead of results.

PurioChat Data Training tab showing the Start Training button and content source cards

Heads up: Embeddings are tied to the AI provider and embedding model you chose. If you switch providers or change the embedding model later, your old embeddings no longer match, so you’ll need to retrain.

What you need to remember

  • Embeddings capture meaning, so PurioChat understands intent rather than matching exact words.
  • Training creates them. No training means no embeddings, which means no AI search or context for the chatbot.
  • Better-trained content means better answers. The more relevant content you train, the more the AI has to draw on.

That’s the whole idea. Next, point PurioChat at your content and let it learn.