Embedding

An embedding is the translation of a text into a vector of numbers that represents its meaning. Texts with similar content sit close together in this numeric space, regardless of the exact words. “How do I cancel my subscription?” and “end my contract” produce similar embeddings even though they share barely a word.

Why embeddings matter

Embeddings make semantic search possible: instead of searching for exact keywords, the system finds passages that match in meaning — even when users phrase things quite differently from your documents. This is exactly the basis of the retrieval step in RAG: the question is turned into an embedding and compared with the embeddings of your content to find the most relevant passages.

Embeddings at Kyros

When you add content to your knowledge base — via URL crawl, document upload or golden answer — it is stored as embeddings in the background. When someone later asks a question, the assistant uses embeddings to find the matching passages and composes a grounded, sourced answer from them. You don’t have to deal with this technique — it runs automatically.