tigergraph/graphrag repository overview

The TigerGraph GraphRAG Docker image offers a streamlined, containerized deployment of an advanced AI assistant designed to bridge the gap between your enterprise data and Large Language Models (LLMs).

A major advantage of this architecture is its consolidated backend: the core TigerGraph database functions simultaneously as both the graph database and the vector database. By natively handling both structural data connections and semantic vector embeddings, TigerGraph serves as the single, unified resource for all of GraphRAG's operations. This eliminates the operational overhead of synchronizing disparate vector stores and graph databases.

Drawing from the project's high-level architecture, the Docker image spins up an AI assistant with two primary components running securely on top of this unified database:

• Natural Language Assistant: Converts user questions into actionable data queries, executing hybrid retrievals (combining vector search with multi-hop graph traversals) to ground LLMs in factual, context-rich data.
• Knowledge Graph Builder: A document ingestion engine that automatically extracts entities, builds ontologies, and structures raw data into an intelligent knowledge graph.

Ready to deploy? You can deploy the Docker-based instance in just a few steps. Head over to the official repository to get started: 👉 https://github.com/tigergraph/graphrag⁠

For a comprehensive overview, prerequisites, and high-level architectural information, please reference the official README.md⁠.