In depth
Embeddings turn text into geometry. Two pieces of text with similar meaning land near each other in vector space, regardless of whether they share words. This is what makes semantic search possible: 'how do I cancel my subscription' and 'where do I unsubscribe' produce nearby vectors. Embedding models are themselves LLMs, just smaller and tuned for the embedding task.
Related concepts
RAGA pattern where relevant documents are fetched from a knowledge base and pasted into context before the LLM answers.Vector databaseA database optimised for storing and querying high-dimensional vectors, typically for similarity search.Semantic searchSearching by meaning instead of keywords. Powered by embeddings and vector databases.
Newsletter
Get the next post in your inbox.
Engineering notes from the Digitorn team. No marketing, no launch announcements, no "10 prompts that will change your life". Just the things we write that we'd want to read.