All Tools
A
RAGFreeOpen Source
AUTOFLOW
Graph RAG knowledge base with conversational search and citations
Apache-2.0
ABOUT
Standard RAG pipelines retrieve flat chunks of text based on embedding similarity, missing the relationships between entities, documents, and concepts that give context to answers. Autoflow builds a knowledge graph from ingested documents using LLM-extracted entities and relations, then uses Graph RAG — combining graph traversal with vector search — to answer questions with context-aware citations. It integrates DSPy for automatic optimization of retrieval and generation prompts, producing higher-quality answers than flat retrieval alone.
INSTALL
docker compose upINTEGRATION GUIDE
1. Build a knowledge base that answers questions with citations sourced from a graph of interconnected documents
2. Create a conversational search engine for internal documentation with entity-aware context and follow-up support
3. Deploy a Perplexity-style internal search tool that crawls websites and builds a queryable knowledge graph
4. Optimize RAG pipelines automatically with DSPy-driven prompt tuning for better retrieval and generation quality
TAGS
ragknowledge-graphgraphragvector-searchconversationaldspy