Vector DBs & RAG
Vector databases and RAG infrastructure give LLMs a memory beyond their context window — storing embeddings, running similarity search, and retrieving the right chunks of your own data so answers are grounded instead of guessed.
-
HaystackOpen-source Python framework from deepset for building production RAG pipelines and agents, with a modular pipeline architecture that predates the current agent framework wave.
-
MilvusOpen-source vector database built for billion-scale similarity search, with a distributed, cloud-native architecture and a managed cloud option called Zilliz Cloud.
-
LanceDBEmbedded, serverless vector database built on the Lance columnar format, designed to run without a separate server process for multimodal AI and agent memory.
-
pgvectorOpen-source Postgres extension that adds vector similarity search directly into an existing database, letting teams skip a separate vector database entirely.
-
WiziOpen-source, self-hosted natural-language code search for React codebases, built on OpenAI embeddings and Pinecone. Labeled a prototype.
-
turbopuffer
Serverless vector and full-text search database built on object storage, designed for cheap search over very large datasets.
-
Vespa
Open-source search and recommendation engine combining vector, lexical and structured search with on-node ML inference, usable self-hosted or as Vespa Cloud.
-
PotpieOpen-source platform that turns a codebase, its history, tickets, and docs into a knowledge graph, with pre-built agents for debugging, testing, and review.
-
LlamaIndex
Open-source data framework for RAG and agents over enterprise data, with 150-plus connectors through LlamaHub for sources like Notion and SQL.
-
Pinecone
Managed, serverless vector database for production search and agentic retrieval, with hosted embedding and reranking models through Pinecone Inference.
-
Weaviate
Open-source vector database with native hybrid search that fuses BM25 keyword matching and vector similarity in a single query.
-
Chroma
Open-source embedding database built to be the simplest way to add vector search to a Python or JavaScript app, from local prototypes to production.
-
Qdrant
Open-source vector database written in Rust, built for fast filtered search with hybrid dense and sparse retrieval.