RAG Overview

​

Overview

​

Which RAG Should I Use?

​

Use AutoRAG if:

• ✅ You want the easiest setup
• ✅ Your documents are in R2
• ✅ You want automatic updates
• ✅ You don’t need custom chunking logic

​

Use Vectorize-RAG if:

• ✅ You need custom chunking strategies
• ✅ You want manual control over embeddings
• ✅ You need specific metadata schemas
• ✅ You’re building complex RAG pipelines

​

Use Generic RAG if:

• ✅ You’re using a non-Cloudflare vector database
• ✅ You need maximum flexibility
• ✅ You have existing RAG infrastructure

​

Quick Comparison

​

Examples

​

AutoRAG (Easiest)

- member: search
  type: AutoRAG
  config:
    instance: my-docs
    mode: answer
  input:
    query: ${input.question}

​

Vectorize-RAG (Flexible)

- member: search
  type: VectorizeRAG
  config:
    vectorizeBinding: VECTORIZE
    indexName: custom-index
    operation: query
  input:
    query: ${input.question}
    topK: 5
    filters:
      category: ${input.category}

​

Generic RAG (Universal)

- member: search
  type: RAG
  config:
    provider: pinecone
    apiKey: ${env.PINECONE_KEY}
  input:
    query: ${input.question}

​

Get Started