Embedding
An embedding is a dense numerical vector that represents a piece of content (text, image, audio) in a way that semantically similar items end up close together in the vector space.
What is Embedding?
Embeddings are produced by neural networks trained for similarity. A sentence like 'reset my password' will have a vector very close to 'change my login credentials' even though no words overlap. Embeddings power semantic search, clustering, deduplication, recommendation, and the retrieval step of RAG pipelines. Quality and dimensionality vary by model. Embeddings of personal data are themselves personal data under GDPR.
How does Embedding apply to enterprise AI?
Enterprises store document embeddings in a vector database to enable semantic retrieval over policy libraries, knowledge bases, ticket archives, and contract repositories.
Related terms
- Vector Database - A vector database is a storage system optimised for indexing and querying high-dimensional embedding vectors using approximate nearest neighbour search.
- RAG (Retrieval-Augmented Generation) - Retrieval-Augmented Generation (RAG) is an architecture pattern that grounds a language model's output in retrieved source documents rather than relying on the model's parametric memory alone.
- Data Residency - Data residency is the requirement that personal or regulated data stays within a specified geographic region throughout processing, storage, and backup.
External references
Need help applying Embedding to your enterprise? Submit a short brief and we reply within one business day.