Vector Database is a database designed to index, store, and query vector representations of data so that similar items can be found using distance or similarity measures.

Key Points

Vector databases store embeddings rather than only exact keyword matches.
They enable semantic retrieval based on similarity.
They are commonly used in RAG pipelines and AI assistants.
Retrieval speed and index quality affect answer quality and latency.
They are especially useful when the knowledge source is large or unstructured.

Concept

A vector database supports semantic search by comparing embeddings rather than relying only on literal text matching. That makes it useful for finding the most relevant document passages, knowledge articles, or record fragments for an AI model to use as context. In a retrieval pipeline, it usually sits between the user's query and the language model, turning a search problem into a relevance-ranking problem.

Explainer

For ConnectedEarth, vector databases matter because many operational questions are not answered well by simple keyword search. A field note, procedure, asset history, or incident log may need semantic retrieval to surface the right context for an AI agent or operator assistant. The vector database is what makes retrieval augmented generation practical at scale, especially where the relevant information is distributed across large, messy corpora.

Cite this entry

Vector Database. ConnectedEarth Knowledge Base. https://connectedearth.space/knowledge-base/vector-database/ (updated 13 September 2026).