A vector database is a database designed to store and search vectors. In this context, a vector is a list of numbers that represents the meaning or features of something, such as a paragraph, image, product, song, or support ticket. Machine learning models create these vectors so similar things end up close together in mathematical space.
Traditional search often looks for matching words. A vector database can search for similarity instead. If you search for "small laptop for travel," it may find products described as lightweight notebooks even if the exact words are different. That makes vector search useful for recommendations, semantic search, image matching, and AI assistants.
Vector databases are often used with Retrieval-Augmented Generation. In that setup, documents are turned into vectors and stored. When a user asks a question, the system finds the most relevant chunks and gives them to an AI model as context. This helps the model answer using the organization's own information.
A vector database usually includes indexing methods that make similarity search fast. Without an index, comparing a query vector to millions of stored vectors would be slow. The database may also store metadata, such as dates, categories, permissions, or source links, so results can be filtered.
The key point is that a vector database is not just a normal table with numbers. It is built for finding things that are close in meaning. As AI tools become more common, vector databases have become a practical way to connect language models with real content.