RAG · Search

It builds a persistent structural map of your codebase so AI assistants read only the files in a change's blast radius instead of the whole repo.

It exists because pure vector search often misses the big picture, so LightRAG auto-maps entities and relationships into a knowledge graph for richer retrieval.

Graphiti continuously ingests structured and unstructured data into temporal context graphs so agents can query what is true now—or was true at any specific moment—without rebuilding the graph from scratch.

PageIndex replaces vector similarity search with hierarchical tree reasoning to retrieve complex documents more accurately.

Official Jupyter notebooks demonstrating how to wire Claude into production tasks like RAG, SQL queries, and multimodal pipelines.

It catalogs 42 advanced RAG techniques as runnable notebooks so developers can compare chunking, reranking, and graph retrieval strategies without wiring up every pipeline from scratch.

GraphRAG exists to give LLMs a structured memory layer for reasoning over messy, private narrative text.

Because stitching together LLM workflows, RAG, agents, and observability by hand is a full-time job.

Quivr wraps retrieval-augmented generation in a Python library with YAML-configurable workflows, letting you swap LLMs and vector stores without rewriting plumbing.

Kotaemon is an open-source document chat tool built on Gradio that exposes citations, scores, and PDF highlights instead of hiding them.

Langflow is a visual schematic editor for AI workflows that deploys your graphs as APIs or MCP servers, saving you from writing yet another Python script to chain LLM calls together.

RAGFlow fuses deep document parsing with agentic workflows so LLMs can answer from messy corporate documents without making things up.

MemPalace is a local-first AI memory system that stores conversations verbatim and retrieves them with semantic search backed by published, reproducible benchmarks.

GBrain gives AI agents a persistent memory, synthesizing cited answers from your notes and wiring a knowledge graph autonomously while you sleep.

STORM simulates expert research conversations so LLMs can write long, cited articles from scratch.

It centralizes model definitions so the same architecture works across PyTorch, JAX, vLLM, and llama.cpp without rewrites.

A privacy-focused AI answering engine that runs entirely on your own hardware so your searches never leave the building.

Milvus is a Go/C++ vector database built for billion-scale ANN search with Kubernetes-native horizontal scaling.

A maintainer cataloged every Chinese NLP repo they touched into a single, obsessively categorized list so others wouldn’t have to hunt.

An open-source framework built to answer one question: how do you get LLMs to reason over your private PDFs, APIs, and SQL databases?