Vector database
A vector database is a data-management system that stores and indexes numerical vector representations and supports retrieval through similarity or distance searches.
Grokking AI Applications
A vector store is a specialized database that stores text chunks or other items together with vector representations for similarity-based retrieval.
- How it is used
- Supports RAG by storing document knowledge for similarity retrieval before an LLM generates an answer.
- How deep it goes
- Explains ingestion, embeddings, nearest-chunk retrieval, runtime stages, and tuning parameters without database implementation internals.
- Examples
- Uses a restaurant PDF assistant and a cookbook-searching chef agent to demonstrate stored retrieval knowledge.
- What you can do after
- Build it from scratch as an ingestion and retrieve-and-answer flow using Langflow components.
Read how Grokking AI Applications explains vector database →
Generative AI in Action
A vector database is a data-management system that stores and provides access to data together with numerical vector representations and supports vector-based retrieval.
- How it is used
- Supports retrieval of relevant proprietary content for RAG answers without reprocessing the full collection for every question.
- How deep it goes
- Defines storage and indexing roles, then walks through local ingestion and retrieval; leaves production clustering and operations beyond the example.
- Examples
- Uses Redis Stack for chat over blog posts and Chroma with Semantic Kernel for question answering over PDFs.
- What you can do after
- Build it from scratch as a local Redis-backed RAG retrieval workflow, while recognizing additional production engineering needs.
AI Agents and Applications
A vector store is a database specialized for storing embeddings and associated documents and retrieving them by semantic similarity.
- How it is used
- Supports semantic retrieval of document chunks for RAG-based question answering and chatbots.
- How deep it goes
- Defines similarity search, contrasts database and library capabilities, and walks through ingestion and querying; it does not detail indexing algorithms.
- Examples
- Uses Chroma to ingest Wikipedia, Word, PDF, and text content, then inspects similarity-search results and source metadata.
- What you can do after
- Apply it via Chroma and LangChain to store document chunks, run similarity searches, and inspect retrieval before answer generation.
Read how AI Agents and Applications explains vector database →
Spring AI in Action
A vector store is a data store that indexes numerical vector representations and supports similarity-based retrieval of associated items such as documents.
- How it is used
- Supports RAG applications by retrieving relevant document chunks for inclusion in prompts.
- How deep it goes
- Explains chunk loading, embeddings, similarity search, metadata filtering, deployment, and verification, while deferring embedding mathematics.
- Examples
- Uses board-game rulebooks, a Spring Cloud Function pipeline, Qdrant collections, and filtered game-rules queries.
- What you can do after
- Build it from scratch with Spring AI, Docker Compose, Qdrant, document readers, splitters, and VectorStore searches.
Build Python Web Apps with Streamlit
A vector database is a data-storage system optimized for storing, indexing, and searching numerical vector representations by similarity.
- How it is used
- Supplies relevant private support-article passages to a customer-service chatbot through RAG.
- How deep it goes
- Develops similarity through a movie analogy and walks through index configuration, ingestion, and retrieval without requiring database internals.
- Examples
- Uses movie ratings to explain vectors and fountain-pen support articles to demonstrate an indexed chatbot knowledge base.
- What you can do after
- Apply it via Pinecone and LangChain to ingest text articles, retrieve similar chunks, and supply context to a chatbot.
Read how Build Python Web Apps with Streamlit explains vector database →
Effective Conversational AI
A vector store is a storage system that indexes numerical vector representations for similarity search.
- How it is used
- Retrieves relevant document passages by meaning to supply context for a RAG chatbot’s answer.
- How deep it goes
- Explains chunk indexing and query-time similarity retrieval, then gives selection and evaluation criteria without detailing index internals.
- Examples
- Uses a medication question, a Chroma document-ingestion example, and a customer-support retrieval scenario.
- What you can do after
- Evaluate/understand a vector store’s fit using retrieval relevance, recall, latency, indexing demands, scale, and cost.
Read how Effective Conversational AI explains vector database →
Data Pipelines with Apache Airflow, Second Edition
A vector database is a database that stores vector representations and supports operations such as similarity search and vector-based retrieval.
- How it is used
- Supports recipe retrieval for RAG answers while Airflow maintains the underlying collection.
- How deep it goes
- Explains embedding-based retrieval and works through collection configuration, record updates, and task ordering; leaves embedding-model training aside.
- Examples
- Uses family recipes, a lasagna question, a flavor-space analogy, a Weaviate ingestion DAG, and a recipe chat assistant.
- What you can do after
- Build it from scratch using Airflow tasks to populate Weaviate, maintain recipe records, and retrieve context for generated answers.
Read how Data Pipelines with Apache Airflow, Second Edition explains vector database →
LLMs in Production
A vector database is a database optimized for storing and retrieving numerical vector representations using similarity-based search.
- How it is used
- Supplies similarity-based retrieval of embeddings for LLM infrastructure and a coding copilot’s RAG system.
- How deep it goes
- Defines vector search and works through Milvus ingestion, indexing, and querying, without constructing a database engine.
- Examples
- Uses reverse image search and a Milvus collection of Python coding tasks queried with an MNIST classification request.
- What you can do after
- Apply it via Milvus to ingest embedded coding examples and retrieve similar instructions for a RAG application.
Natural Language Processing in Action, Second Edition
A vector store is a software component that stores and retrieves numerical vectors, typically with similarity-search indexes.
- How it is used
- Powers scalable semantic search and retrieval-augmented generation over document collections.
- How deep it goes
- Explains embeddings, similarity ranking, ANN indexing, storage, updates, and deployment choices without detailing every indexing algorithm.
- Examples
- Names Milvus, Weaviate, Qdrant, pgvector, OpenSearch, Pinecone, Haystack, and Knowt as implementation or pipeline examples.
- What you can do after
- Build semantic-search and RAG pipelines by selecting embeddings, ANN indexes, vector stores, and retrieval frameworks.
Read how Natural Language Processing in Action, Second Edition explains vector database →
MongoDB 8.0 in Action, Third Edition
A vector database is a database system that stores, retrieves, and searches numerical vectors by their similarity in a multidimensional space.
- How it is used
- Supports similarity-based retrieval for semantic search, recommendations, and retrieval-augmented generation.
- How deep it goes
- Explains embedding-to-query workflow and Atlas search parameters, but defers a working embedding-generation query example.
- Examples
- Uses a “village life” movie-plot search and a chatbot retrieving prevalidated MongoDB documentation from a PDF.
- What you can do after
- Apply it via Atlas Vector Search to retrieve similar documents and via LangChain to supply context to a RAG chatbot.
Read how MongoDB 8.0 in Action, Third Edition explains vector database →
Machine Learning Platform Engineering
A vector database is a database system that stores numerical vector representations and supports similarity-based retrieval.
- How it is used
- Supplies meaning-based document retrieval for RAG, providing relevant context before LLM generation.
- How deep it goes
- Walks through embedding, indexing, and query retrieval with FAISS; stops short of vector-index algorithm internals.
- Examples
- Uses API-credential questions, a FAISS document store, and manual versus LangChain retrieval code.
- What you can do after
- Apply it via FAISS and LangChain to retrieve document context for a RAG application.
Read how Machine Learning Platform Engineering explains vector database →
Build AI-Enhanced Web Apps
A vector store is a database or storage abstraction that indexes high-dimensional vectors for efficient similarity-based retrieval.
- How it is used
- Supports semantic document retrieval for web search and RAG answers grounded in uploaded knowledge-base content.
- How deep it goes
- Defines vector stores and walks through ingestion, similarity search, and persistent indexing without implementing a search algorithm.
- Examples
- Uses a MemoryVectorStore text search and a PDF-to-HNSWLib indexing routine.
- What you can do after
- Build it from scratch using LangChain.js vector-store APIs to index documents, retrieve matches, and supply context to RAG.
Read how Build AI-Enhanced Web Apps explains vector database →
Building Reliable AI Systems
A vector database is a database designed to store and retrieve vector representations, often by similarity.
- How it is used
- Stores embedded document chunks for similarity retrieval and metadata-filtered context selection in RAG.
- How deep it goes
- Explains indexes, metadata, and production capabilities, then walks through Pinecone setup and queries without implementing an indexing algorithm.
- Examples
- Builds a health-care policy assistant with chunk metadata, medical-domain embeddings, a Pinecone index, and filtered queries.
- What you can do after
- Apply it via Pinecone to index policy chunks and retrieve filtered context for a RAG answer.
Read how Building Reliable AI Systems explains vector database →
Domain-Specific Small Language Models
A database designed to store vectors and support similarity search, typically using approximate nearest neighbor methods for retrieval.
- How it is used
- Supports RAG systems by storing embeddings and retrieving semantically similar context for language-model generation.
- How deep it goes
- Explains storage, approximate search, indexing, schemas, and retrieval, but defers detailed ANN algorithm internals.
- Examples
- Uses LanceDB tables, cosine IVF_FLAT indexing, and a "blocked KV-cache" query returning matching document passages.
- What you can do after
- Operate/configure LanceDB to persist embeddings, define a schema, build an index, and retrieve matching records.
Read how Domain-Specific Small Language Models explains vector database →
AI Agents in Action, Second Edition
A vector database server is a service that exposes storage and similarity retrieval over vector representations.
- How it is used
- Stores document-chunk vectors for similarity retrieval that supplies context to RAG answers and agent memory.
- How deep it goes
- Defines similarity lookup and works through an in-memory prototype and Chroma-backed retrieval, without detailing database indexing internals.
- Examples
- Uses a scored TF-IDF document search and a movie-script agent backed by a persistent Chroma DB collection.
- What you can do after
- Build it from scratch as an in-memory similarity search, then apply Chroma DB to retrieve embedded document chunks.
Read how AI Agents in Action, Second Edition explains vector database →
Investing for Programmers
A vector database is a database designed to store and retrieve vector representations of data, typically for similarity-based search.
- How it is used
- Supports retrieval-augmented generation by storing embedded document chunks for prompt-specific context retrieval.
- How deep it goes
- Explains chunking, embedding, insertion, and similarity search, but defers production database management.
- Examples
- Uses text files, overlapping document chunks, OpenAI embeddings, an in-memory store, and autonomous-driving company questions.
- What you can do after
- Apply it via LangChain by embedding documents, storing them, retrieving similar context, and passing results to an LLM.
Read how Investing for Programmers explains vector database →
Introduction to Generative AI, Second Edition
A vector database is a data-management system optimized for storing and searching high-dimensional vector representations by similarity.
- How it is used
- Supports retrieval-augmented generation by finding relevant external information before a language model generates a response.
- How deep it goes
- Explains embeddings, semantic similarity, nearest-neighbor retrieval, and context insertion, but does not detail database implementation.
- Examples
- Uses travel-policy retrieval and vegan-restaurant recommendations involving user memory, Yelp, and web search.
- What you can do after
- Apply it via retrieval-augmented generation to ground responses in related documents and make supporting sources inspectable.
Read how Introduction to Generative AI, Second Edition explains vector database →
Data Storytelling with Altair and AI
A vector database is a database optimized for storing and retrieving numerical vector representations by similarity or other vector operations.
- How it is used
- Extends an LLM with domain-specific context for RAG answers, commentaries, and annotations.
- How deep it goes
- Walks through loading HTML, chunking text, generating embeddings, persisting Chroma data, and querying it through RetrievalQA.
- Examples
- Uses a New SmartX 2023 HTML product page, a Chroma collection, and a 30-word product-description query.
- What you can do after
- Apply it via Chroma and LangChain to build a retriever-backed LLM question-answering application.
Read how Data Storytelling with Altair and AI explains vector database →
The Art of AI Product Development
A vector database is a database optimized for storing numeric vector representations and retrieving items by vector similarity.
- How it is used
- Supports semantic search by storing document embeddings and retrieving items according to vector similarity.
- How deep it goes
- Explains distributional similarity and the storage-and-retrieval workflow, without detailing database internals.
- Examples
- Uses apple, banana, and car to illustrate distances between semantically similar and dissimilar embeddings.
- What you can do after
- Apply it via semantic search to retrieve documents whose meanings resemble a query.
Read how The Art of AI Product Development explains vector database →
Just Use Postgres!
A vector database is a database designed to store vector representations and support efficient similarity searches over them.
- How it is used
- Supports recommendation services, private-data retrieval, RAG applications, and autonomous AI agents.
- How deep it goes
- Explains storage, similarity search, embeddings, and specialized indexes, while deferring detailed index operation and deployment guidance.
- Examples
- Uses a movie recommendation service, private datasets, Docker commands, and pgvector extension setup as concrete examples.
- What you can do after
- Operate/configure Postgres with pgvector and use stored embeddings for similarity search and RAG context retrieval.
Transformers in Action
A vector database is a database optimized for storing, indexing, and searching vector representations by their similarity.
- How it is used
- Deploys as the retrieval store in RAG systems that supply language models with relevant external context.
- How deep it goes
- Explains storage, embeddings, similarity retrieval, approximate nearest-neighbor search, and tuning concerns, but stops short of database internals.
- Examples
- Uses chunked documents, query embeddings, top-matching passages, and Flashrank or JinaRerank refinement as concrete pipeline examples.
- What you can do after
- Operate/configure RAG retrieval by selecting a vector store, embedding model, retrieval strategy, chunk size, and top-k value.
AI-Powered Search
A vector database is a database optimized for storing, indexing, and retrieving vector representations by similarity.
- How it is used
- Supports retrieval-augmented generation by supplying language models with accurate, up-to-date information as context.
- How deep it goes
- Explains the search and grounding role, but stops short of indexing, similarity metrics, storage, or operational procedures.
- Examples
- Uses retrieval-augmented generation and a convergence diagram spanning textual, audio, video, metadata, sensory, and temporal modalities.
- What you can do after
- Apply it via retrieval-augmented generation as a knowledge source for grounding language-model responses.
AI Agents in Action
A vector database is a database optimized for storing numerical vector representations and retrieving records by vector-based similarity or distance.
- How it is used
- Supports RAG, document question answering, semantic search, and agent memory by retrieving relevant stored content.
- How deep it goes
- Explains indexing, embedding, querying, similarity scores, and cosine distance, while deferring nuanced retrieval design.
- Examples
- Demonstrates an in-memory Python search database and a local Chroma DB collection queried with document embeddings.
- What you can do after
- Build and query a simple vector store in Python or Chroma DB for document retrieval.
Software Testing with Generative AI
A vector database is a database specialized for storing and retrieving numerical vector representations using similarity-based search.
- How it is used
- Supports retrieval-augmented generation by selecting contextual data for prompts.
- How deep it goes
- Defines similarity-based storage and retrieval, but defers database internals and algorithm details.
- Examples
- Uses querying a database to pull multiple relevant files into a prompt.
- What you can do after
- Operate vector databases in RAG frameworks to retrieve contextual files for prompts.
Read how Software Testing with Generative AI explains vector database →
Building LLM Applications with DSPy
A vector store is a data system that indexes and retrieves numerical vector representations, typically by similarity.
- How it is used
- Uses vector similarity to retrieve relevant documents for a RAG system.
- How deep it goes
- Explains embedding, storage, top-k retrieval, and a corpus-size-based backend choice without detailing database internals.
- Examples
- Uses a discount-pricing question to retrieve matching article titles, then applies the approach to a knowledge base.
- What you can do after
- Apply it via DSPy’s Embeddings retriever to return the top-k similar items from a corpus.
Read how Building LLM Applications with DSPy explains vector database →
Build Applications with Local AI Models on a Mac
A vector database stores vector representations and supports searching for vectors that are similar or close to a query vector.
- How it is used
- Supports RAG by providing access to a user's documents through vector search.
- How deep it goes
- Defines vector storage and similarity search, but does not explain implementation or internals.
- Examples
- Names ChromaDB as an example of a vector database used for RAG.
- What you can do after
- Understand vector databases as a component of RAG architectures.
Read how Build Applications with Local AI Models on a Mac explains vector database →
Related concepts
- Chroma DBincludes · Chroma is an open-source vector database for storing and querying embeddings and associated data, such as documents and metadata.
- Pineconeincludes · Pinecone is a managed cloud vector database service for storing, indexing, and querying vector embeddings, including through similarity search.
- VectorStoreChatMemoryAdvisorused by · VectorStoreChatMemoryAdvisor is a Spring AI advisor that stores chat messages as embeddings in a vector store and retrieves relevant messages as conversational context.
- retrieval-augmented generationused by · Retrieval-augmented generation is a technique that retrieves relevant information from external sources and supplies it as context to a generative model when producing an output.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.