RAG Retriever
Retrieves relevant documents from a knowledge base using vector similarity search to augment LLM context.
How It Works
1Receives a query string from the input
2Converts the query to an embedding vector
3Searches the vector store for the top-K most similar documents
4Returns the retrieved documents as context for the LLM
Configuration
| Parameter | Description | Default |
|---|---|---|
| knowledge_base | Select from your organization's knowledge bases | - |
| top_k | Number of documents to retrieve | 4 |
| similarity_threshold | Minimum similarity score (0.0-1.0) | 0.7 |
| embedding_model | Model used for query embedding | - |
| metadata_filters | Filter search by document metadata | - |
Supported Vector Stores
Chroma
Built-in, no external setup needed
PostgreSQL pgvector
Use existing PostgreSQL with pgvector extension
Pinecone
Managed vector database service
Weaviate
Open-source vector search engine