What is a vector database?
A vector database stores data as embeddings, lists of numbers that capture meaning, and finds the items closest to a query, so a search for ‘refund’ also finds ‘money back’.
Why it matters for a business
If you plan an AI assistant over your documents, the vector store decides how well it finds the right passage.
More
It is the usual storage layer behind RAG and AI search.
You may not need a separate product: PostgreSQL with the pgvector extension covers many projects, while dedicated vector databases suit very large collections. The choice affects hosting, cost and where your data lives.
Embeddings
An embedding model turns each text chunk into a vector; similar meanings end up close together.
How search works
The query is turned into a vector too, and the database returns the nearest stored vectors, often combined with keyword filters.
Keep it in sync
When a document changes or is deleted, its vectors must change too, or the AI will quote old text.
- InputA question, a message, a document
- vector databaseContextYour data, found by search
- Language modelReads, reasons and writes
- ResultAn answer, or an action in your tools
What it costs
Starting prices of our services on this topic, in your currency.
Vector database in practice
A database that stores text or images as vectors so it can find items by meaning, not exact words.
Example
- A law firm in Oxford wants staff to search years of advice notes by meaning.
- Each document is split into passages, turned into embeddings and stored in a vector database, with client and matter as filters.
- A question about a lease break clause then finds relevant passages even when they use different words.
Which term is it?
A large language model: an AI model trained on huge amounts of text to read and write language.

From the founder
I want AI that does a real job inside your tools, with a person checking what matters. We test it on your own data before you rely on it.
Anatoly MikhailovFounder of Pazl







