---
language: en
title: "Vector database"
description: "A vector database stores and queries vectors by proximity. Learn how to evaluate filters, updates, and retrieval accuracy in your application."
canonical: "https://tropk.ai/en/resources/glossary/vector-database"
last_modified: "2026-10-12T16:13:10.306Z"
---

# Vector database

Reviewed on 2026-10-06.

## What it means

A vector database is a system that stores vectors and lets you query them by proximity, using a distance or similarity measure. In applications involving text, these vectors are usually [embeddings](/en/resources/glossary/embeddings) associated with documents and metadata. This capability may be available in a dedicated service or as an extension to a conventional database, such as pgvector for PostgreSQL.

## Why it matters for marketing

For marketing, the database influences which content an assistant can retrieve from a knowledge base and how quickly. An evaluation should consider updates, filters, and candidate quality, as well as the advertised scale. Having materials stored does not guarantee that the system will return the current version or respect criteria such as country, category, and access permissions.

## Example in practice

Imagine a distributor that keeps Brazilian and Portuguese catalogs in the same vector database. A query about delivery terms finds a semantically similar document, but it is intended for Portugal. An inspection reveals that the country was not included in the filter. The team decides to require the Brazilian criteria in the query and test again before revising the catalog copy.

## How to apply or verify

Ask for a demonstrable query with vectors, source documents, and recorded filters. Confirm how additions, changes, and deletions reach the database. Compare approximate search with an exact reference on a sample, and record missed candidates and response time. In pgvector, approximate indexes may leave out some nearby candidates to return results more quickly; also evaluate how they interact with filters.

## Important distinctions

A vector database does not necessarily generate embeddings or write the final answer. Nor does it need to be a product separate from the relational database. Exact search finds the neighbors defined by distance; approximate search may leave some out to gain speed. Even the mathematically closest neighbor may be unsuitable for the business, requiring additional selection and review criteria.

## Sources and references

- [pgvector: Open-source vector similarity search for Postgres](<https://github.com/pgvector/pgvector>) — Projeto pgvector
- [Vector embeddings](<https://platform.openai.com/docs/guides/embeddings>) — OpenAI

## Continue a leitura

- [Semantic search](</en/resources/glossary/semantic-search>)
- [Embeddings](</en/resources/glossary/embeddings>)
- [RAG](</en/resources/glossary/rag>)

## About the author

[Vinicius Pacheco](</en/team/vinicius-pacheco>)

Responsible for Tropk’s technical content, focusing on crawling, indexing, structured data, and agent access to content.

---

[All pages in Markdown](<https://tropk.ai/sitemap.md>)