Flexible GraphRAG v0.5.0 now supports RDF-based ontologies for both property graph databases and RDF triple store databases (Graphwise Ontotext GraphDB, Fuseki, and Oxigraph). Document ingestion with KG extraction, auto incremental data source change detection, and UI search (hybrid search, AI query, and AI chat) are all supported with both database types. Also added OpenAI-Like, OpenRouter, LiteLLM, and vLLM LLM providers. For RDF search, install with the additional LangChain option: uv pip install flexible-graphrag[langchain] Flexible GraphRAG github repo: https://github.com/stevereiner/flexible-graphrag
Flexible GraphRAG now also supports ArcadeDB embedded in addition remote ArcadeDB. Both work with the optional incremental update auto sync from data sources. All supported databases (10 vector, 3 search, and 8 graph databases) now work with auto sync. Github https://github.com/stevereiner/flexible-graphrag. With a full hybrid setup (vector, search, graph) all get auto synced. Finally have pypi built releases (0.4.0) of flexible-graphrag https://pypi.org/project/flexible-graphrag/ and flexible-graphrag-mcp https://pypi.org/project/flexible-graphrag-mcp/
Big News: Flexible GraphRAG now supports Incremental Update Auto Sync from Data Sources Huge New Feature: Flexible GraphRAG check-in: Added Incremental Update Auto Sync from 9 data sources to all supported vector, search, and graph databases. Added Postgres state management. 🌟 Github: https://github.com/stevereiner/flexible-graphrag
• All auto syncing data sources working: S3 (SQS events), Hyland Alfresco (ActiveMQ events), Azure Blob, GCS (Pub/Sub), Google Drive, Box, OneDrive, SharePoint, Filesystem (watchdog).
• All 10 vector databases working with this (Qdrant, LanceDB, Elastic, OpenSearch Project, Chroma, Milvus, created by Zilliz, Weaviate, Pinecone, PostgreSQL pgvector, Neo4j)
• All 3 full text search engines working with this (Elasticsearch, OpenSearch, built-in LlamaIndex BM25)
• Graph databases working with this: Neo4j, FalkorDB, ArcadeDB (fix to delete coming), MemGraph. More testing needed with Kuzu, NebulaGraph, Amazon Neptunes.
Flexible GraphRAG or Flexible RAG, an Apache 2.0 open source Python platform, uses LlamaIndex , supports 8 graph databases, 10 vector databases, 3 search engines, and 13 data sources (9 auto-sync) Also supports knowledge graph auto-building, schemas, 9 LlamaIndex LLMs, Docling or LlamaParse doc processing, GraphRAG mode, RAG only mode, Hybrid search, AI query, AI chat. Has React, Vue, and Angular TypeScript frontends, a FastAPI backend, and a REST API. Also has a FastMCP MCP server. Has a convenient docker compose that can include any of the databases (Vector, Graph, Search, Alfresco) and Dashboards / Consoles. Blog: https://integratedsemantics.org/ X.com: https://x.com/stevereiner
Shown: updated Flexible GraphRAG with Neo4j, Observability with Grafana Labs , GitHub 🌟 https://github.com/stevereiner/flexible-graphrag , updates: 1. Observability support added with OpenTelemetry, Prometheus, Grafana, Jaeger, and OpenLIT 2. SharePoint and OneDrive data sources fixed and now working (13 data sources: filesystem, Alfresco, SharePoint, Box, CMIS, S3, GCS, Azure Blob, OneDrive, Google Drive, web sources) 3. Added LLM Providers (now 9 total) : OpenAI, Azure OpenAI, Google Gemini, Vertex AI, Anthropic Claude, Amazon Bedrock, Ollama, Grog, Fireworks AI, and Ollama. Uses LlamaIndex. 4. Embedding providers can be used independently.. 5. Like KG Spaces (Github 🌟 https://github.com/stevereiner/kg-spaces-aca) for Hyland Alfresco ACA, added a “Skip graph” checkbox for per ingest control. Skip graph allows on each ingest to have fast RAG ingest speed (vector+full-text) or a full GraphRAG ingest (vector+full-text+graph). Hybrid search and AI query / AI chat works with all that is available from mixed kinds of ingests. 6. Flexible GraphRAG MCP server: added support for all 13 data sources, per ingest skip_graph option , data source config options, full REST API parity. 7. Backend REST API (also supports all 13 data sources), added support for per ingest skip graph, also for KG Spaces: multi-file ingest NodeDetails list with nodeids, get by path for regular Flexible GraphRAG with updated Python Alfresco API package (Github 🌟https://github.com/stevereiner/python-alfresco-api) 8. Added Python backend pyproject.toml, llama-core (no llama-index) and integrations only, to avoid package version issues.
Flexible GraphRAG or Flexible RAG, an Apache 2.0 open source Python platform, uses LlamaIndex , supports 8 graph databases, 10 vector databases, 3 search engines, and 13 data sources, Also supports knowledge graph auto-building, schemas, 9 LlamaIndex LLMs, Docling or LlamaParse doc processing, GraphRAG mode, RAG only mode, Hybrid search, AI query, AI chat. Has React, Vue, and Angular frontends, and a FastAPI backend. Also has a FastMCP MCP server. Has a convenient docker compose that can include any of the databases (Vector, Graph, Search, Alfresco) and Dashboards / Consoles.
Multi select files / folders in regular ACA UI, in process tab can optionally choose skip graph ( search + vector only). Hybrid search tab (with hybrid search or AI query), and AI chat tab (Search is OpenSearch Project or Elastic, vector is one of 10 vector databases, graph is one of 8 graph databases including Neo4j, ArcadeDB, FalkorDB).
Flexible GraphRAG or Flexible RAG, an Apache 2.0 open source python platform, uses LlamaIndex , supports 8 graph databases, 10 vector databases, 3 search engines, and 13 data sources, Also supports knowledge graph auto-building, schemas, LlamaIndex LLMs, Docling and LlamaParse doc processing, GraphRAG mode, RAG only mode, Hybrid search, and AI query / chat. Has React, Vue, and Angular frontends, and a FastAPI backend. React, Vue, Angular, Also has a FastMCP MCP server.
