Skip to main content

Installation

Install VectorAI DB with Docker. Takes about two minutes.

Quickstart

Create a collection, insert vectors, and run your first similarity search.

Core concepts

Understand the data model, architecture, and how search works.

Academy

Hands-on tutorials for semantic search, RAG, hybrid search, and more.

​
New to VectorAI DB?

Follow this path to get up and running.
1

Install

Set up VectorAI DB locally using Docker.
2

Quickstart

Run the Quickstart to create a collection, insert vectors, and run a similarity search.
3

Learn the core concepts

Read the Overview to understand the data model, architecture, and how a search works. It links into the full reference from there.
4

Build something

Try the Academy tutorials for hands-on walkthroughs, or connect to LangChain or LlamaIndex.

​
Use cases

Use caseDescriptionReal-world example
Semantic searchFind documents, products, or records by meaning rather than exact keywords. Store embeddings from any model and query them with a vector.A major e-commerce platform processes 216M+ embeddings daily so shoppers can find products by intent, not exact keywords, across millions of storefronts.
Hybrid searchCombine dense vector similarity with sparse or keyword scoring using fusion strategies such as RRF or DBSF.A global travel platform activated over a billion reviews with hybrid retrieval to power its AI trip planner, driving 2–3x more revenue per engaged user.
RAG pipelinesUse VectorAI DB as the retrieval layer in LangChain, LlamaIndex, or a custom pipeline. Retrieve the most relevant context chunks before sending to a language model.A conversational AI platform retrieves from 100M+ vectors across 12,000+ isolated namespaces to power personalized AI agents in under a second.
Agent memoryGive AI agents persistent, queryable memory. Store past interactions as vectors and retrieve semantically relevant history at runtime.A food delivery platform’s AI shopping agents use a persistent memory layer that lifted grocery-checkout conversion ~24% and cut intent misreads ~33%.
Air-gapped / edge deploymentRun fully on-premises with no outbound network requirements. Suitable for secure or regulated environments.A defense contractor’s AI platform lets U.S. government agencies fine-tune models on classified infrastructure with zero external network connections.