{
  "id": "knowledge-fabric-impl:langchain-rag-fabric",
  "_kind": "KnowledgeFabricImpl",
  "_file": "agent-stack/knowledge-fabric-impls/rag-framework-fabrics.yaml",
  "_cluster": "agent-stack",
  "attributes": {
    "displayName": "LangChain as RAG Knowledge Fabric",
    "description": "LangChain as a RAG-based knowledge fabric implementation. Provides the\nfull pipeline from document loading (100+ loaders for PDFs, web pages,\ndatabases, APIs) through text splitting, embedding, vector storage, and\nretrieval chain construction. LangChain is not a knowledge store itself\nbut an orchestration layer that turns any combination of document sources\nand vector stores into a queryable knowledge fabric. Retrieval chains\ncompose retriever + LLM into a question-answering system grounded in\norganizational knowledge.\n",
    "knowledgeFileFormats": [
      "any (via document loaders)"
    ],
    "retrievalStrategy": "hybrid",
    "knowledgePersistence": "delegated (via vector store backends)",
    "knowledgeScopes": [
      "project",
      "organization",
      "enterprise"
    ],
    "autoExtractionSupport": false,
    "notes": "LangChain's strength as a knowledge fabric is flexibility — it can\ncompose any combination of document sources, embedding models, vector\nstores, and retrieval strategies. The LangChain Expression Language (LCEL)\nenables declarative retrieval chain construction. The trade-off is\ncomplexity — LangChain adds abstraction layers that can obscure the\nunderlying retrieval behavior and make debugging harder.\n"
  },
  "outgoingEdges": [
    {
      "from": "knowledge-fabric-impl:langchain-rag-fabric",
      "to": "layer:12-knowledge-fabric",
      "kind": "realizes",
      "attributes": {}
    },
    {
      "from": "knowledge-fabric-impl:langchain-rag-fabric",
      "to": "framework:langchain",
      "kind": "integrates_with",
      "attributes": {}
    }
  ],
  "incomingEdges": []
}